WO2021135653A1

WO2021135653A1 - Method and system for identifying abnormal stay of vehicle

Info

Publication number: WO2021135653A1
Application number: PCT/CN2020/127694
Authority: WO
Inventors: 张腾剑; 陈奥
Original assignee: 北京嘀嘀无限科技发展有限公司
Priority date: 2019-12-31
Filing date: 2020-11-10
Publication date: 2021-07-08
Also published as: CN111860561A; CN111860561B

Abstract

A method for identifying abnormal stay of a vehicle (150). The method comprises the following steps: obtaining real-time stay data of the vehicle (150) (501); and determining, on the basis of the real-time stay data, whether the vehicle (150) is in an abnormal stay state using an identification model (503), wherein the identification model comprises a machine learning model obtained by training historical stay data of a plurality of historical vehicles (150).

Description

Method and system for identifying abnormal stay of vehicles

Priority statement

This application claims the priority of the Chinese patent application with application number 201911421142.4 filed on December 31, 2019, which is incorporated herein by reference in its entirety.

Technical field

This application relates to the field of traffic technology, and in particular to a method and system for identifying abnormal stops of vehicles.

Background technique

With the development of network technology and communication technology, various transportation services such as online car-hailing, takeaway and other vehicle services have become more and more popular. For the vehicle service platform, safety is the guarantee for the sustainable development of the platform. Abnormal stays in the itinerary are the most common and accounted for the largest proportion of all safety scenarios. The vast majority of cases are caused by driver and passenger conflicts or traffic accidents that cause injuries and deaths. Based on the safety considerations of drivers and passengers, real-time monitoring of whether the vehicle has abnormal stops during the journey can provide safety protection for vehicle-related personnel such as drivers and/or passengers. Therefore, it is necessary to propose a method and system for determining abnormal parking behavior of vehicles.

Summary of the invention

Some embodiments of the application provide a system. The system includes at least one storage medium that includes instructions for identifying an abnormal stay of the vehicle. And at least one processor, the at least one processor is in communication with the at least one storage medium, wherein, when the instruction is executed, the at least one processor is configured to perform the following operations. Acquire real-time stay data of the vehicle; based on the real-time stay data, use a recognition model to determine whether the vehicle is in an abnormal stay state; wherein the recognition model includes machine learning trained based on historical stay data of multiple historical vehicles model.

Some embodiments of the present application provide a method for identifying an abnormal stop of a vehicle. The method includes the following operations. Acquire real-time stay data of the vehicle; based on the real-time stay data, use a recognition model to determine whether the vehicle is in an abnormal stay state; wherein the recognition model includes machine learning trained based on historical stay data of multiple historical vehicles model.

Some embodiments of the present application provide a computer-readable storage medium that stores computer instructions. After the computer reads the computer instructions in the storage medium, the computer executes the following method for identifying abnormal parking of a vehicle. The method includes acquiring real-time stay data of the vehicle; based on the real-time stay data, using a recognition model to determine whether the vehicle is in an abnormal stay state; wherein the recognition model includes training based on historical stay data of a plurality of historical vehicles The resulting machine learning model.

Some embodiments of the present application provide a method for recognizing abnormal stay behavior of online car-hailing. The method includes the following operations. Collect real-time data when the ride-hailing takes place, the real-time data includes the current stay location, current stay time, driver service status, and predetermined characteristic information; obtain the first historical data set and the second historical data set from the database; The first historical data set includes the historical stay time of at least one driver in the same service state as the real-time data, and the second historical data set includes the history of at least one driver in the same time period and in the same area as the real-time data. Staying time; according to the real-time data, the first historical data set, the second historical data set, and a preset model, identify whether the online car-hailing currently exhibits an abnormal staying behavior.

Some embodiments of the present application provide a method for training an abnormal stay behavior recognition model for online car-hailing. The method includes the following operations. Obtain a training data set from a database. The training data includes positive training data belonging to abnormal staying behaviors and negative training data not belonging to abnormal staying behaviors; for each training data, the staying position and staying duration of the training data are extracted , Driver service status and predetermined feature information; obtain a first historical data set and a second historical data set from the database; the first historical data set includes the history of at least one driver in the same service state as the training data Staying time, the second historical data set includes the historical staying time of at least one driver in the same time period and in the same area as the training data; according to the training data set, the first historical data corresponding to each training data The data set and the second historical data set train the preset model so that the accuracy and/or recall rate of the abnormal stay behavior judgment result output by the preset model reaches the target value.

Some embodiments of the present application provide a device for identifying abnormal stay behavior of online car-hailing. The device includes an acquisition module, an acquisition module, and a processing module. The collection module is used to collect real-time data when the online car-hailing stops, and the real-time data includes the current staying position, the current staying time, the driver's service status, and predetermined characteristic information. The acquiring module is used to acquire a first historical data set and a second historical data set from a database; the first historical data set includes the historical stay time of at least one driver in the same service state as the real-time data, the The second historical data set includes the historical stay time of at least one driver in the same time period and in the same area as the real-time data. The processing module is configured to identify whether the online car-hailing currently exhibits abnormal stay behavior based on the real-time data, the first historical data set, the second historical data set, and a preset model.

Some embodiments of the present application provide a device for training an abnormal stay behavior recognition model for online car-hailing. The device includes an acquisition module, a feature extraction module and a training module. The acquisition module is used to acquire a training data set from a database, and the training data includes positive training data belonging to abnormal staying behaviors and negative training data not belonging to abnormal staying behaviors. The feature extraction module is used for extracting the stay location, stay duration, driver service status and predetermined feature information of the training data for each training data. The acquiring module is further configured to acquire a first historical data set and a second historical data set from the database; the first historical data set includes the historical stay time of at least one driver in the same service state as the training data The second historical data set includes the historical stay time of at least one driver in the same time period and in the same area as the training data. The training module is configured to train a preset model according to the training data set, the first historical data set and the second historical data set corresponding to each training data, so that the output of the preset model is abnormal The accuracy and/or recall rate of the judgment result of the staying behavior reaches the target value.

Some embodiments of the present application provide a device for identifying abnormal stay behavior of online car-hailing. The device includes a memory, a processor, and a computer program. Wherein, the computer program is stored in the memory and is configured to be executed by the processor to implement the above-mentioned method for identifying abnormal stay behavior of online car-hailing.

Some embodiments of the present application provide a training device for a recognition model of abnormal stay behavior of online car-hailing. The device includes a memory, a processor, and a computer program. Wherein, the computer program is stored in the memory, and is configured to be executed by the processor to realize the above-mentioned training method of the abnormal stay behavior recognition model of online car-hailing.

Description of the drawings

This application will be further described in the form of exemplary embodiments, and these exemplary embodiments will be described in detail with the accompanying drawings. These embodiments are not restrictive. In these embodiments, the same number represents the same structure, in which:

Fig. 1 is a schematic diagram of an exemplary vehicle monitoring system according to some embodiments of the present application;

Fig. 2 is a schematic diagram of exemplary hardware components and/or software components of an exemplary computing device according to some embodiments of the present application;

Fig. 3 is a schematic diagram of exemplary hardware components and/or software components of an exemplary mobile device according to some embodiments of the present application;

Fig. 4 is a block diagram of an exemplary processing device according to some embodiments of the present application;

Fig. 5 is an exemplary flowchart of a method for identifying abnormal stay behavior of a vehicle according to some embodiments of the present application;

Fig. 6 is an exemplary flow chart for determining whether a vehicle is in an abnormal stay state by using a recognition model according to some embodiments of the present application;

FIG. 7 is an exemplary flowchart for determining whether the current stay duration is an abnormal stay duration according to some embodiments of the present application;

FIG. 8 is an exemplary flowchart for determining whether the characteristic information related to the current vehicle stay meets the stay rule according to some embodiments of the present application; and

Fig. 9 is an exemplary flowchart of a method for training a recognition model according to some embodiments of the present application.

Detailed ways

In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings that need to be used in the description of the embodiments. Obviously, the drawings in the following description are only some examples or embodiments of the application. For those of ordinary skill in the art, without creative work, the application can be applied to the application according to these drawings. Other similar scenarios. Unless it is obvious from the language environment or otherwise stated, the same reference numerals in the figures represent the same structure or operation.

It should be understood that the “system”, “device”, “unit” and/or “module” used herein is a method for distinguishing different components, elements, parts, parts, or assemblies of different levels. However, if other words can achieve the same purpose, the words can be replaced by other expressions.

As shown in the present application and claims, unless the context clearly suggests exceptional circumstances, the words "a", "an", "an" and/or "the" do not specifically refer to the singular, but may also include the plural. Generally speaking, the terms "include" and "include" only suggest that the clearly identified steps and elements are included, and these steps and elements do not constitute an exclusive list, and the method or device may also include other steps or elements.

In this application, a flowchart is used to illustrate the operations performed by the system according to the embodiment of the application. It should be understood that the preceding or following operations are not necessarily performed exactly in order. Instead, the steps can be processed in reverse order or at the same time. At the same time, you can also add other operations to these processes, or remove a step or several operations from these processes.

The embodiments of the present application can be applied to different transportation systems, and different transportation systems include, but are not limited to, one or a combination of land, ocean, aviation, aerospace, and the like. For example, taxis, private cars, ride-hailing cars, buses, agent driving, trains, high-speed trains, ships, airplanes, hot air balloons, unmanned vehicles, collection/delivery express delivery, etc. which apply management and/or distribution transportation systems . The application scenarios of different embodiments of the present application include, but are not limited to, one or a combination of web pages, browser plug-ins, clients, customized systems, enterprise internal analysis systems, artificial intelligence robots, and the like. It should be understood that the application scenarios of the system and method of the present application are only some examples or embodiments of the present application. For those of ordinary skill in the art, they can also be based on these drawings without creative work. Apply this application to other similar scenarios. For example, other similar guidance users to park systems.

The "passenger", "passenger terminal", "user terminal", "customer", "demand", "service demander", "consumer", "consumer", "user demander", etc. described in this application are Interchangeable refers to the party that needs or subscribes to the service. It can be an individual or a tool. Similarly, the “driver”, “driver’s end”, “provider”, “provider”, “service provider”, “service provider”, “service provider”, etc. described in this application are also interchangeable and refer to Individuals, tools, or other entities that provide services or assist in providing services. In addition, the "user" described in this application may be a party that needs or subscribes to services, or a party that provides services or assists in providing services.

Fig. 1 is a schematic diagram of an exemplary vehicle monitoring system according to some embodiments of the present application. In some embodiments, the vehicle monitoring system 100 can be applied to a variety of application scenarios. For example, the vehicle monitoring system 100 can be applied to a variety of transportation services. The transportation services can include taxi services, express services, private car services, minibus services, carpooling services, ride-hailing services, bus services, driving services, courier services, and takeaways. Services, freight services, etc.

In some embodiments, the vehicle monitoring system 100 may determine whether the stay of the vehicle performing the service is an abnormal stay. For example, for a ride-hailing service, the vehicle monitoring system 100 may determine that the service vehicle is in a service state (for example, in the process of picking up passengers) and/or in a non-service state (for example, in the process of waiting for a service request) for one or more stops. Whether it is in an abnormal state. For another example, for a take-out service, the vehicle monitoring system 100 can determine whether the take-out person's vehicle is in an abnormal stop state during one or several stops in the way of picking up and/or delivering the meal. For another example, for freight services, the vehicle monitoring system 100 can determine whether the delivery vehicle is in an abnormal stay state during one or more stops during the freight transportation.

As shown in FIG. 1, the vehicle monitoring system 100 may include a processing device 110, a terminal 120, a storage device 130, a network 140 and a vehicle 150. The processing device 110 may process data and/or information obtained from the terminal 120, the storage device 130, and/or the vehicle 150. In some embodiments, the processing device 110 may obtain location/trajectory information of multiple terminals 120 and/or characteristic information of persons related to the vehicle (for example, drivers and passengers). The processing device 110 may process the information and/or data obtained above to perform one or more functions described in this application. For example, the processing device 110 may collect real-time stay data of the vehicle 150 and store it in the storage device 130, and may also obtain historical stay data from the storage device 130 to train the recognition model. In the process of identifying abnormal parking behaviors of the vehicle, the processing device 110 can identify whether the vehicle 150 currently has abnormal parking behaviors based on the collected real-time stay data when the vehicle 150 stays, and according to the recognition model trained from the historical stay data . In some embodiments, the processing device 110 may be an independent server or a server group. The server group may be centralized or distributed (for example, the processing device 110 may be a distributed system). In some embodiments, the processing device 110 may be local or remote. For example, the processing device 110 may access and obtain information and/or data from the terminal 120 and/or the storage device 130 and/or the vehicle 150 through the network 140. In some embodiments, the processing device 110 may be directly connected to the terminal 120, the storage device 130, and/or the vehicle 150 to obtain information and/or data. In some embodiments, the processing device 110 may be executed on a cloud platform. For example, the cloud platform may include one or any combination of private cloud, public cloud, hybrid cloud, community cloud, decentralized cloud, internal cloud, etc. In some embodiments, the processing device 110 may be integrated in the terminal 120. In some embodiments, the processing device 110 may be implemented by the computing device 200 shown in FIG. 2.

In some embodiments, the processing device 110 may include one or more sub-processing devices (for example, a single-core processor or a multi-core processor). For example only, the processing device 110 may include a central processing unit (CPU), an application specific integrated circuit (ASIC), an application specific instruction processor (ASIP), a graphics processing unit (GPU), a physical processor (PPU), a digital signal processor ( DSP), Field Programmable Gate Array (FPGA), Editable Logic Circuit (PLD), Controller, Microcontroller Unit, Reduced Instruction Set Computer (RISC), Microprocessor, etc. or any combination thereof.

The terminal 120 may be a device with data acquisition, storage and/or transmission functions. The terminal 120 may include a service provider terminal, a service requester terminal, a vehicle-mounted terminal, and the like. Service providers can be individuals, tools, or other entities that provide services. Service requesters can be individuals, tools, or other entities that need to get or are receiving services. For example, for a taxi service, the service provider may be a driver or a third-party platform, and the service requester may be a passenger or other person or device (for example, an Internet of Things device) that receives similar services. In some embodiments, the terminal 120 may be used to collect various types of data, including but not limited to service-related data. The data collected by the terminal 120 may be real-time data or various historical data. In some embodiments, the terminal 120 may collect data through its own sensors, acquire data from external sensors, or read data from its own memory, or read data from the storage device 130 through the network 140. In some embodiments, the sensors of the terminal 120 may include positioning devices, sound sensors, image sensors, temperature and humidity sensors, position sensors, pressure sensors, distance sensors, speed sensors, acceleration sensors, gravity sensors, displacement sensors, torque sensors, and gyroscopes.仪, etc. or any combination thereof.

In some embodiments, the terminal 120 may include a mobile device 120-1, a vehicle built-in device 120-2, a tablet computer 120-3, a notebook computer 120-4, a desktop computer 120-5, etc., or any combination thereof. In some embodiments, the mobile device 120-1 may include a smart home device, a wearable device, a smart mobile device, a virtual reality device, an augmented reality device, etc., or any combination thereof. Smart home devices may include smart lighting devices, smart electrical control devices, smart monitoring devices, smart TVs, smart cameras, walkie-talkies, etc., or any combination thereof. Wearable devices may include smart bracelets, smart footwear, smart glasses, smart helmets, smart watches, smart clothes, smart backpacks, smart accessories, etc., or any combination thereof. Smart mobile devices may include smart phones, personal digital assistants (PDAs), game devices, navigation devices, POS machines, etc., or any combination thereof. The virtual reality/augmented reality device may include a virtual reality helmet, virtual reality glasses, virtual reality goggles, augmented reality helmets, augmented reality glasses, augmented reality goggles, etc., or any combination thereof. In some embodiments, the in-vehicle device 120-2 may include an in-vehicle computer, an in-vehicle data recorder, an in-vehicle human-computer interaction (HCI) system, a driving recorder, an in-vehicle TV, and the like.

In some embodiments, the terminal 120 may be a device with a positioning function, and its positioning function may be realized by a variety of positioning technologies, for example, the global positioning system (GPS), the global satellite navigation system (GLONASS), and the Beidou navigation system (BDS) , Galileo positioning system (GALILEO), quasi-zenith satellite system (QZSS), wireless fidelity (Wi-Fi) positioning technology, etc. In some embodiments, the terminal 120 may transmit the collected data/information to the processing device 110 via the network 140 for subsequent steps. The terminal 120 may also store the collected data/information in its own memory, or transmit it to the storage device 130 via the network 140 for storage. In some embodiments, the vehicle monitoring system 100 may include a plurality of interconnected terminals to collect various data together, and one or more of the terminals may preprocess the data.

The storage device 130 may store data and/or instructions. In some embodiments, the storage device 130 may store data/information acquired by the terminal 120 and/or the processing device 110. For example, the storage device 130 may store historical service data related to historical transportation services performed by multiple vehicles (for example, historical service order data, historical service participant data, historical vehicle-related data, historical travel data, historical order evaluation data, etc.). In some embodiments, the storage device 130 may store data and/or instructions used by the processing device 110 to execute or use to complete the exemplary methods described in this application. In some embodiments, the storage device 130 may be a part of the processing device 110 or the terminal 120.

In some embodiments, the storage device 130 may include mass memory, removable memory, volatile read-write memory, read-only memory (ROM), etc., or any combination thereof. Exemplary mass storage devices may include magnetic disks, optical disks, solid state disks, and the like. Exemplary removable storage may include flash drives, floppy disks, optical disks, memory cards, compact disks, magnetic tapes, and the like. An exemplary volatile read-only memory may include random access memory (RAM). Exemplary RAM may include dynamic RAM (DRAM), double rate synchronous dynamic RAM (DDR SDRAM), static RAM (SRAM), thyristor RAM (T-RAM), zero capacitance RAM (Z-RAM), and the like. Exemplary ROMs may include mask ROM (MROM), programmable ROM (PROM), erasable programmable ROM (EPROM), electronically erasable programmable ROM (EEPROM), compact disk ROM (CD-ROM), and digital General-purpose disk ROM, etc. In some embodiments, the storage device 130 may be implemented on a cloud platform. For example only, the cloud platform may include private cloud, public cloud, hybrid cloud, community cloud, distributed cloud, internal cloud, multi-layer cloud, etc., or any combination thereof. In some embodiments, specific types of historical data may be uniformly stored on a cloud platform, so that multiple processing devices 110 or terminals 120 can access or update, and ensure the real-time and cross-platform use of the data.

In some embodiments, the storage device 130 may be connected to the network 140 to communicate with one or more components in the vehicle monitoring system 100 (for example, the processing device 110, the terminal 120, the vehicle 150). One or more components in the vehicle monitoring system 100 can access data or instructions stored in the storage device 130 through the network 140. In some embodiments, the storage device 130 may directly connect or communicate with one or more components in the vehicle monitoring system 100 (for example, the processing device 110, the terminal 120, and the vehicle 150).

The network 140 may facilitate the exchange of information and/or data. In some embodiments, one or more components in the vehicle monitoring system 100 (for example, the processing device 110, the terminal 120, the storage device 130, and the vehicle 150) may send and receive data to/from other components in the vehicle monitoring system 100 through the network 140. /Or receive information and/or data. For example, the processing device 110 may obtain data/information related to the current transportation service from the terminal 120 and/or the vehicle 150 through the network 140. For another example, the processing device 110 may also obtain historical service data related to historical transportation services from the storage device 130 via the network 140.

In some embodiments, the network 140 may be any form of wired or wireless network or any combination thereof. For example only, the network 140 may include a cable network, a wired network, an optical fiber network, a telecommunication network, an internal network, the Internet, a local area network (LAN), a wide area network (WAN), a wireless local area network (WLAN), a metropolitan area network (MAN), public switched telephone network (PSTN), Bluetooth ^(TM) network, ZigBee ^TM network, near field communication (NFC) network, global system for mobile communications (GSM) network, Code division Multiple Access (CDMA) networks, Time division Multiple Access (TDMA) network , General Packet Radio Service (GPRS) network, Enhanced Data Rate GSM Evolution (EDGE) Network, Wideband Code Division Multiple Access (WCDMA) Network, High Speed Downlink Packet Access (HSDPA) Network, Long Term Evolution (LTE) Network, User Datagram Protocol (UDP) network, Transmission Control Protocol/Internet Protocol (TCP/IP) network, short message service (SMS) network, wireless application protocol (WAP) network, ultra-wideband (UWB) network, mobile communication (1G, 2G) , 3G, 4G, 5G) networks, Wi-Fi, Li-Fi, Narrowband Internet of Things (NB-IoT), etc. or any combination thereof. In some embodiments, the vehicle monitoring system 100 may include one or more network access points. For example, the vehicle monitoring system 100 may include wired or wireless network access points, such as base stations and/or wireless access points, through which one or more components of the vehicle monitoring system 100 may be connected to the network 140 to exchange data and/or information.

The vehicle 150 may be a carrier for performing service requests. For example, the vehicle 150 may be used to provide a taxi service, and the driver uses the vehicle 150 to pick up passengers and then deliver the passengers to the destination. In some embodiments, the vehicle 150 may include an in-vehicle terminal. In some embodiments, relevant personnel of the vehicle 150 (for example, a driver and/or passengers) may be equipped with a user terminal. The type of the vehicle-mounted terminal and/or the user terminal and the terminal 120 may be the same or different. In some embodiments, the vehicle 150 may obtain driving data generated by itself during driving, for example, speed, acceleration, and so on. The above driving data can be transmitted to other components of the vehicle monitoring system 100, for example, the processing device 110 through the network 120. In some embodiments, the vehicle 150 may be an electric vehicle, a fuel cell vehicle, a hybrid vehicle, or a traditional internal combustion engine vehicle. The body type of the vehicle 150 may be any body type, for example, a motorcycle, a sports car, a coupe, a sedan, a pickup truck, a station wagon, a sports utility vehicle (SUV), a minivan, or a transfer car.

A person of ordinary skill in the art should understand that when a component of the vehicle monitoring system 100 is executed, the component may be executed by an electrical signal and/or an electromagnetic signal. For example, when the terminal 120 processes a task, such as determining, identifying, or selecting an object, the terminal 120 may operate a logic circuit in its processor to process the task. When the terminal 120 sends a service request to the processing device 110, the processor of the terminal 120 may generate an electrical signal encoding the request. Then, the processor of the terminal 120 may send the electrical signal to the output port. If the terminal 120 communicates with the processing device 110 via a wired network, the output port may be physically connected to a cable, which further transmits electrical signals to the input port of the processing device 110. If the terminal 120 communicates with the processing device 110 through a wireless network, the output port of the terminal 120 may be one or more antennas, and the antennas may convert electrical signals into electromagnetic signals. In an electronic device, when the processor of the electronic device processes instructions, sends instructions, and/or performs actions, the instructions and/or actions are conducted via electrical signals. For example, when the processor retrieves or saves data from a storage medium (for example, the storage device 130), it can send an electrical signal to a read/write device of the storage medium, and the read/write device can read or write in the storage medium. Structured data. The structured data can be transmitted to the processor via the bus of the electronic device in the form of electrical signals. Here, the electrical signal may refer to one electrical signal, a series of electrical signals, and/or multiple discontinuous electrical signals.

In order to solve the problems of real-time and low recall rate in the existing technology through the driver, passengers, and police actively reporting to the online car-hailing platform, and then the identification method of the abnormal parking behavior of the online car-hailing, and the passage and fixed length of stay The method for identifying abnormal staying behaviors based on threshold comparison has the problem of low accuracy. The embodiment of the present invention provides a method for identifying abnormal staying behaviors of online car-hailing, which can collect real-time data when online car-hailing occurs, including current Multi-dimensional feature information such as stay location, current stay time, driver service status, and scheduled feature information, combined with multi-dimensional related historical data in the database, and the intelligent recognition model obtained by pre-training, can identify the network in real time and accurately. Whether there are abnormal stay behaviors in the current ride-hailing, it can avoid misjudgment caused by the fixed stay time threshold, and improve the real-time performance, accuracy and recall rate, thereby improving the safety perception ability of the online ride-hailing platform and the ability to detect traffic accidents and drivers. The ability to recognize passenger conflicts ensures the safety of drivers and passengers.

The method for identifying abnormal stay behaviors of online car-hailing provided by the embodiments of the present application may be applicable to communication systems. The communication system includes a server of a car-hailing platform, a database, and a car-hailing terminal (including a vehicle-mounted terminal, a driver terminal, or a passenger terminal) connected in communication with the server. The server can be a server, or a server cluster composed of several servers, or a cloud computing service center, which can collect real-time data of online car-hailing from the online car-hailing terminal and store it in the database, or it can be obtained from the database historical data. In the process of identifying the abnormal stay behavior of online car-hailing, the server can obtain relevant historical data from the database according to the collected real-time data when the car-hailing occurs, and according to the real-time data, historical data and predetermined model, It can be identified whether the online car-hailing currently exhibits abnormal staying behavior.

In this application, the online car-hailing service platform can be the same as or similar to the transportation service platform, the server of the online car-hailing platform can be the same as or similar to the processing device 110, and the online car-hailing terminal can be the same as the terminal 120. Same or similar.

Fig. 2 is a schematic diagram of exemplary hardware components and/or software components of an exemplary computing device according to some embodiments of the present application. In some embodiments, the processing device 110 and/or the terminal 120 may be implemented by the computing device 200. As shown in FIG. 2, the computing device 200 may include a processor 210, a memory 220, an input/output (I/O) 230, and a communication port 240.

The processor 210 may execute computer instructions (for example, program code) and may perform the functions of the processing device 110 according to the technology described in this application. Computer instructions can be used to perform specific functions described in this application, and computer instructions can include programs, objects, components, data structures, programs, modules, functions, and so on. In some embodiments, the processor 210 may include one or more hardware processors, such as a microcontroller, a microprocessor, a reduced instruction set computer (RISC), and an application specific integrated circuit. circuit (ASIC)), application-specific instruction-set processor (ASIP), central processing unit (CPU), graphics processing unit (GPU)) , Physical processing unit (physics processing unit (PPU)), digital signal processor (digital signal processor (DSP)), field programmable gate array (FPGA)), advanced RISC machine (advanced RISC machine( ARM)), programmable logic device (PLD), any circuit or processor capable of performing one or more functions, or a combination of several of them.

For illustration only, only one processor is described in the computing device 200. However, it should be noted that the computing device 200 may also include multiple processors. The operations and/or methods described in this application that are executed by one processor may also be executed jointly or separately by multiple processors. For example, if the processor of the computing device 200 described in this application performs operation A and operation B, it should be understood that operation A and operation B can also be shared by two or more different processors in the computing device 200. Or separately (for example, the first processor performs operation A and the second processor performs operation B, or the first processor and the second processor jointly perform operations A and B).

The memory 220 may store data/information acquired from the processing device 110, the terminal 120, the storage device 130, and/or any other components of the vehicle monitoring system 100. In some embodiments, the memory 220 may include mass memory, removable memory, volatile read-write memory, read-only memory (ROM), etc., or any combination thereof. Mass storage can include magnetic disks, optical disks, solid state drives, and mobile storage. Removable storage may include flash drives, floppy disks, optical disks, memory cards, ZIP disks, tapes, and so on. Volatile read-write memory may include random access memory (RAM). RAM can include dynamic random access memory (DRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), static random access memory (SRAM), thyristor random access memory (t-ram), zero capacitance random access memory Take memory (Z-RAM) and so on. ROM can include mask read-only memory (MROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), optical disk read-only memory Storage (CD-ROM), digital versatile disc, etc. In some embodiments, the memory 220 may store one or more programs and/or instructions for executing the exemplary methods described in this application.

The input/output 230 may input and/or output signals, data, information, and the like. In some embodiments, the input/output 230 may implement the interaction between the user and the processing device 110. In some embodiments, the input/output 230 may include an input device and an output device. The input device may include a keyboard, a mouse, a touch screen, a microphone, etc., or any combination thereof. The output device may include a display device, a speaker, a printer, a projector, etc. or any combination thereof. The display device may include a liquid crystal display (LCD), a light emitting diode (LED) display, a flat panel display, a curved screen, a television device, a cathode ray tube (CRT), a touch screen, etc., or any combination thereof.

The communication port 240 may be connected to a network (for example, the network 140) to facilitate data communication. The communication port 240 may establish a connection between the processing device 110 and the terminal 120, the storage device 130, and/or the vehicle 150. The connection may be a wired connection, a wireless connection, any connection capable of data transmission and/or reception, etc., or any combination thereof. Wired connections can include cables, optical cables, telephone lines, etc., or any combination thereof. The wireless connection may include a Bluetooth ^™ link, a Wi-Fi ^™ link, a WiMAX ^™ link, a wireless local area network link, a ZigBee ^™ link, a mobile network link (for example, 3G, 4G, 5G, etc.), etc., or any combination thereof. In some embodiments, the communication port 240 may be and/or include a standardized communication port, such as RS232, RS485, and the like.

Fig. 3 is a schematic diagram of exemplary hardware components and/or software components of an exemplary mobile device according to some embodiments of the present application. In some embodiments, the terminal 120 may be implemented by the mobile device 300. As shown in FIG. 3, the mobile device 300 may include a communication unit 310, a display unit 320, a graphics processing unit (GPU) 330, a central processing unit (CPU) 340, an input/output unit 350, a memory 360, and a storage unit 390. In some embodiments, the mobile device 300 may also include any other suitable components, including but not limited to a system bus or a controller (not shown in the figure). In some embodiments, the mobile operating system 370 (for example, iOS ^TM , Android, Windows Phone ^TM, etc.) and one or more application programs 380 can be loaded from the storage unit 390 into the memory 360 so as to be executed by the central processing unit 340 . The application program 380 may include a browser or any other suitable mobile application program for receiving and presenting information from the processing device 110. The user interaction of the information flow may be implemented through the input/output unit 350 and provided to the processing device 110 and/or other components of the vehicle monitoring system 100 through the network 140.

In order to realize the various modules, units and functions described in this application, a computer hardware platform may be used as a hardware platform for one or more elements described in this application. A computer with user interface elements can be used to implement a personal computer (PC) or any other type of workstation or terminal device. If properly programmed, the computer can also act as a server.

Fig. 4 is a block diagram of an exemplary processing device according to some embodiments of the present application. As shown in FIG. 4, the processing device 110 may include an acquisition module 410, a determination module 420, and a training module 430.

The obtaining module 410 may be used to obtain real-time stay data of the vehicle. The real-time stay data may include the current stay time and characteristic information related to the stay of the vehicle. In some embodiments, in order to improve the safety of drivers and passengers, the acquisition module 410 may acquire real-time data when the vehicle stops. Specifically, the obtaining module 410 may obtain real-time data from the vehicle 150 or the terminal 140. The real-time data may include, but is not limited to, location information, time information, some status information and interaction information of persons related to the vehicle (for example, drivers and/or passengers). In some embodiments, the processing device 110 (for example, the acquisition module 410) may acquire positioning data of the vehicle, including real-time speed data, and determine whether the vehicle is staying based on the real-time speed data. When the vehicle stays, the acquisition module 410 may acquire the status of the service provider of the current stay of the vehicle and the characteristic information related to the vehicle stay from the above real-time data.

The determination module 420 may use the recognition model to determine whether the vehicle is in an abnormal stay state based on the real-time stay data. In some embodiments, the determination module 420 may use the recognition model to process the real-time stay data to realize the real-time recognition of the abnormal stay behavior of the vehicle. The recognition model may be a fusion model, and may include a first sub-recognition model, a second sub-recognition model, and a third sub-recognition model. The recognition model can judge whether the stay of the vehicle is an abnormal stay behavior from different dimensions. As an example, the first sub-recognition model may be a distribution-based recognition model, the second sub-recognition model may be a statistics-based sub-recognition model, and the third model may be a density-based sub-recognition model.

In some embodiments, the determining module 420 may determine whether the current stay duration is an abnormal stay duration based on the first sub-recognition model. The determining module 420 may input the current stay duration into the first sub-recognition model, and determine whether the current stay duration is an abnormal stay duration. The first sub-recognition model may be any recognition model constructed based on an abnormal value detection algorithm. For example, the outlier detection algorithm may include clustering-based methods, isolated forests, statistical methods, distribution-based Tukey Method and so on. The first sub-recognition model can determine whether the current stay time is an outlier relative to the historical stay time used to train the first sub-recognition model. If yes, it can be determined that the current stay duration is the abnormal stay duration.

In some embodiments, in order to determine whether the current stay duration is an abnormal stay duration based on the first sub-recognition model, the determining module 420 may determine the recognition coefficient based on the distribution-based Dukai method post-comparison model, and based on the recognition The coefficient determines at least one maximum anomaly estimate. In some embodiments, the determination module 420 may determine whether the current stay time is greater than at least one of the at least one maximum abnormality estimation value. In response to the current stay time being greater than at least one of the at least one maximum abnormality estimation value, the determining module 420 may determine that the current stay time length is the abnormal stay time length.

In some embodiments, in response to the current stay duration being the abnormal stay duration, the determining module 420 may determine whether the characteristic information related to the vehicle stay meets the stay rule based on the second sub-recognition model. The second recognition sub-model may be a rule model based on statistics. The determining module 420 may determine whether the characteristic information related to the vehicle stay meets the stay rule determined based on the second recognition sub-model.

In some embodiments, to determine whether the characteristic information related to the vehicle stay meets the stay rule based on the second sub-recognition model, the determining module 420 may determine the stay rule based on the second sub-recognition model. The stay rule may be determined based on the preset training of the second sub-recognition model. For example, by statistically analyzing the historical stay data of multiple vehicles to obtain stay rules. In some embodiments, the determination module 420 may determine whether the characteristic information related to the stay of the vehicle satisfies the stay rule. The characteristic information related to the stay of the vehicle may include at least the behavior of the person related to the vehicle and the current stay position of the vehicle. The behavior of the personnel related to the vehicle may include the behavior of the service requester and/or the service provider. The behavior of the service requester may include payment behavior, evaluation behavior, conflict behavior, etc. of the service requester (for example, a passenger requesting a taxi service). The behavior of the service provider may include the evaluation behavior, service behavior, conflict behavior, etc. of the service provider (for example, a driver who provides a taxi service). The determining module 420 may determine whether the behavior of the person related to the vehicle is a normal behavior, and/or determine whether the current stay location is in a hot spot area. The normal behavior may include payment behavior (for example, active payment), evaluation behavior (for example, active praise), and the like. Abnormal behavior may include conflict behavior (for example, driver and passenger conflict) and the like. The hot spot area may include a service request sending hot spot area and/or a traffic jam area. If the determination module 420 determines that the determination result of at least one of the above items is yes, it can be determined that the characteristic information related to the current vehicle stay meets the stay rules preset by the second model, so as to determine that the vehicle currently does not exhibit abnormal stay behavior. If the above judgment results are all no, it is determined that the characteristic information related to the current vehicle stay does not satisfy the stay rule preset by the second model.

In some embodiments, the characteristic information related to the stay in response to the vehicle does not satisfy the stay rule. The determining module 420 may determine the abnormal evaluation value related to the current stay time in the third sub-recognition model. The determining module 420 may input the current stay time into the third sub-recognition model, and obtain the abnormal evaluation value of the current stay time. The determining module 420 may obtain the abnormal evaluation value of the current stay time relative to the sample data for training the model through the third sub-recognition model. The sample data for training the model may include second historical stay data of at least one second sample vehicle in the same time period and in the same area. The third sub-recognition model may be any model that can determine outliers, for example, Robust Random Cut Forest, Isolation Forest, Elliptic Envelope, OneClassSVM, Gaussian outlier detection, etc. In some embodiments, the third sub-recognition model may include a density-based local outlier factor model (Local Outlier Factor, LOF model). The determining module 420 may obtain the local abnormality factor (LOF value) of the current stay time through the LOF model, and designate the LOF value as the abnormality evaluation value.

In some embodiments, the determination module 420 may determine whether the abnormality evaluation value is greater than the abnormal stay threshold. The determination module 420 may compare the abnormality evaluation value with the abnormal stay threshold value to determine whether the abnormality evaluation value is greater than the abnormal stay threshold value. In response to the abnormality evaluation value being greater than the abnormal stay threshold value, the determination module 420 may determine that the vehicle is in an abnormal stay state.

In some embodiments, when the determining module 420 determines whether the vehicle is in an abnormal stay state, if the current stay time is not the abnormal stay time, or the characteristic information related to the vehicle stay meets the stay rules, determine the vehicle If the vehicle is not in an abnormal parking state, or the abnormal evaluation value is less than or equal to the preset threshold value, the determination module 420 may determine that the vehicle is not in an abnormal parking state.

In some embodiments, the determining module 420 may respectively use the sub-models included in the recognition model to recognize the abnormal stay of the vehicle. As an example, the determining module 420 may determine whether the current stay duration is an abnormal stay duration based on the first sub-recognition model. At the same time, the determining module 420 may determine whether the characteristic information related to the vehicle stay meets the stay rule based on the second sub-recognition model. In addition, the determining module 420 may also determine the abnormal evaluation value related to the current stay time based on the third sub-model and determine whether the abnormal evaluation value is greater than a preset threshold. When the current stay time is the abnormal stay time, the characteristic information related to the vehicle stay does not meet the stay rules, and/or the abnormal evaluation value is greater than the preset threshold, the determining module 420 may determine that the vehicle In an abnormal state.

The training module 430 can train a recognition model. The training module 430 can obtain multiple training samples. The multiple training samples include first historical stay data of at least one first sample vehicle in the same vehicle state, and second historical stay data of at least one second sample vehicle in the same time period and in the same area, including those belonging to Positive samples of abnormal staying behaviors and negative samples that do not belong to abnormal staying behaviors. Each training sample includes the historical stay time of the vehicle, the status of the service provider (such as driving status, arrival status, service status) or non-service status (such as order status, receiving status), and characteristic information related to the current vehicle stay .

In some embodiments, the training module 430 may train the first initial sub-recognition model based on the first historical stay data to obtain the first sub-recognition model and the recognition parameters related to the first sub-recognition model. The first initial sub-recognition model may be any recognition model constructed based on an abnormal value detection algorithm. For example, the outlier detection algorithm may include clustering-based methods, isolated forests, statistical methods, distribution-based Tukey Method and so on. The training module 430 may use the first historical stay data to train the first sub-recognition model while determining the recognition parameters.

In some embodiments, the training module 430 may train the second initial sub-recognition model based on the first historical stay data and the second historical stay data to obtain the second sub-recognition model and stay rules related to the second sub-recognition model. The second initial sub-recognition model may include a rule model based on statistics. The training module 430 may perform statistical analysis on the training samples to construct the second initial sub-model. For example, the training samples whose historical stay duration included in the sample belong to the abnormal stay duration but are negative samples are analyzed to determine one or more preset rules. Subsequently, the final stay rule can be obtained through continuous debugging and/or modification of the preset rule. At the same time, the trained second sub-recognition model can be obtained.

In some embodiments, the training module 430 may train a third initial sub-recognition model based on the second historical stay data to obtain a third sub-recognition model and an abnormal stay threshold related to the third sub-recognition model. The third initial sub-recognition model can be any model that can determine outliers, for example, Robust Random Cut Forest, Isolation Forest, Elliptic Envelope, OneClassSVM, Gaussian outlier detection, and so on. In some embodiments, the third sub-recognition model may include a density-based local outlier factor model (Local Outlier Factor, LOF model). The training module 430 may use the second historical stay data to train the third initial sub-recognition model, for example, continuously adjust the abnormal stay threshold used to evaluate the local abnormality factor (LOF value) to obtain the third sub-recognition model, and the final Abnormal stay threshold.

In some embodiments, two or more modules in the processing device 110 may be combined into a single module, and any one of the modules may be divided into two or more units. For example, the acquisition module 410 and the determination module 420 may be combined into a single module for acquiring real-time stay data and determining whether the vehicle is in an abnormal stay. For another example, the processing device 110 may include a storage module (not shown in FIG. 4), which may be configured to store various data (for example, service requests, historical stay data) involved in the execution of the processing device.

It should be understood that the system and its modules shown in FIG. 4 can be implemented in various ways. For example, in some embodiments, the system and its modules may be implemented by hardware, software, or a combination of software and hardware. Among them, the hardware part can be implemented using dedicated logic; the software part can be stored in a memory and executed by an appropriate instruction execution system, such as a microprocessor or dedicated design hardware. Those skilled in the art can understand that the above-mentioned methods and systems can be implemented using computer-executable instructions and/or included in processor control codes, for example on a carrier medium such as a disk, CD or DVD-ROM, such as a read-only memory (firmware Such codes are provided on a programmable memory or a data carrier such as an optical or electronic signal carrier. The system and its modules of this application can not only be implemented by hardware circuits such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, etc., or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc. It may also be implemented by software executed by various types of processors, or may be implemented by a combination of the above hardware circuit and software (for example, firmware).

In some embodiments of the present application, a device for identifying abnormal stay behavior of online car-hailing is provided, which can execute the processing flow provided in the embodiment of the method for identifying abnormal stay behavior of online car-hailing. The device for identifying abnormal stay behavior of online car-hailing includes a collection module, an acquisition module, and a processing module.

The collection module is used to collect real-time data when a car-hailing stay occurs, and the real-time data includes the current stay location, the current stay time, the driver's service status, and predetermined characteristic information;

The acquiring module is used to acquire a first historical data set and a second historical data set from a database; the first historical data set includes the historical stay time of at least one driver in the same service state as the real-time data, and the first historical data set 2. The historical data set includes the historical stay time of at least one driver in the same time period and in the same area as the real-time data;

The processing module is configured to identify whether the online car-hailing currently exhibits abnormal stay behavior according to the real-time data, the first historical data set, the second historical data set, and a preset model.

On the basis of the foregoing embodiment, the processing module includes:

The first processing module is configured to input the current stay duration and the first historical data set into a first model, and determine whether the current stay duration is abnormal stay duration relative to the first historical data set;

The second processing module is configured to input the predetermined characteristic information into the second model if it is determined that the current stay duration is the abnormal stay duration, and determine whether the predetermined characteristic information meets the preset rules of the second model;

The third processing module is configured to, if it is determined that the predetermined characteristic information does not meet the preset rules of the second model, input the current stay time and the second historical data set into the third model to obtain all The outlier degree factor of the current stay duration, and if the outlier degree factor is greater than a preset threshold, it is determined that the online car-hailing currently has an abnormal stay behavior.

On the basis of the foregoing embodiment, the first model is a distribution-based Tukey Method model, and a coefficient used to represent the degree of abnormality in the Tukey Method model adopts a preset coefficient;

The first processing module is configured to: when determining whether the current stay duration is abnormal stay duration relative to the first historical data set:

Obtaining the maximum estimated value of the first historical data set by using the Tukey Method model with the preset coefficient;

If the current stay duration is greater than the maximum estimated value, it is determined that the current stay duration is the abnormal stay duration.

On the basis of the foregoing embodiment, the first historical data set includes a first subset and a second subset; the first subset includes the current driver's historical stay time in the same service state as the real-time data; The second subset includes the historical stay time of all drivers in the database in the same service state and in the same area as the real-time data;

Obtaining the first maximum estimated value of the first subset and the second maximum estimated value of the second subset using the preset coefficient through the Tukey Method model;

If the current stay duration is greater than the first maximum estimated value or the second maximum estimated value, it is determined that the current stay duration is an abnormal stay duration.

On the basis of the foregoing embodiment, the predetermined characteristic information includes at least one of the following: passenger's payment behavior, passenger's evaluation of the driver, and billing hot spots;

The second processing module is configured to: when determining whether the predetermined feature information satisfies the preset rules of the second model:

Determine whether the payment behavior of the passenger is active payment; and/or

Determine whether the passenger’s evaluation of the driver is positive; and/or

Judging whether the current stay position is in the hot spot area for issuing bills;

If the foregoing judgment results are all no, it is determined that the predetermined characteristic information does not satisfy the preset rule of the second model.

On the basis of the foregoing embodiment, the third model is a density-based LOF model;

When the third processing module obtains the outlier degree factor of the current stay time, it is used to:

According to the LOF model, the outlier degree factor LOF value of the current stay time is obtained according to the current stay time and the second historical data set.

On the basis of the foregoing embodiment, after determining that the predetermined characteristic information does not satisfy the preset rule of the second model, the third processing module is further configured to:

If the number of samples in the second data set is less than the preset number, it is directly determined that the online car-hailing currently has an abnormal staying behavior;

The third processing module is also used for:

The number of historical stay data of all drivers in the database in the same time period as the real-time data is counted, and the result obtained by multiplying the number of historical stay data by a preset percentage is used as the preset number.

On the basis of the foregoing embodiment, the first processing module is further configured to, if it is determined that the current stay duration is not an abnormal stay duration, determine that the online car-hailing does not currently exhibit any abnormal stay behavior; or

The second processing module is further configured to, if it is determined that the predetermined characteristic information satisfies the preset rules of the second model, determine that the online car-hailing does not currently exhibit abnormal stay behavior; or

The third processing module is further configured to, if it is determined that the outlier degree factor is less than or equal to a preset threshold, determine that there is no abnormal staying behavior in the online car-hailing currently.

The device for identifying abnormal stay behavior of online car-hailing provided by the embodiment of the present invention may be specifically used to implement the method embodiments provided in FIGS. 5-8, and the specific functions are not repeated here.

The device for identifying abnormal stay behavior of online car-hailing provided by the embodiment of the present invention collects real-time data when a car-hailing stop occurs, and the real-time data includes current staying position, current staying time, driver service status, and predetermined characteristic information; A first historical data set and a second historical data set are obtained from the database; the first historical data set includes the historical stay time of at least one driver in the same service state as the real-time data, and the second historical data set includes at least The historical stay time of a driver in the same time period and in the same area as the real-time data; according to the real-time data, the first historical data set, the second historical data set, and a preset model, the network is identified Whether there are abnormal stopping behaviors in the current ride-hailing. This embodiment uses massive multi-dimensional historical data and preset models for online car-hailing travel, and takes into account the multi-dimensional feature information such as stay time, location, service status, etc., to accurately identify whether there is any abnormal staying behavior in online car-hailing in real time. , Improve the safety perception ability of the online car-hailing platform and the ability to recognize traffic accidents and driver-passenger conflicts, and ensure the safety of drivers and passengers.

Some embodiments of the present application provide an apparatus for training an abnormal staying behavior recognition model for online car-hailing, which can execute the processing flow provided in the embodiment of the method for training an abnormal staying behavior recognition model for online car-hailing. The device for training an abnormal staying behavior recognition model for online car-hailing includes an acquisition module, a feature extraction module, and a training module.

An obtaining module, configured to obtain a training data set from a database, the training data including positive training data belonging to abnormal staying behaviors and negative training data not belonging to abnormal staying behaviors;

The feature extraction module is used to extract the stay location, stay time, driver service status and predetermined feature information of the training data for each training data;

The acquiring module is further configured to acquire a first historical data set and a second historical data set from the database; the first historical data set includes the historical stay of at least one driver in the same service state as the training data Duration, the second historical data set includes the historical stay duration of at least one driver in the same time period and in the same area as the training data;

The training module is used to train the preset model according to the training data set, the first historical data set and the second historical data set corresponding to each training data, so that the abnormal output of the preset model stays The accuracy and/or recall rate of the behavior judgment result reaches the target value.

On the basis of the foregoing embodiment, the training module is used for:

Input the length of stay and the first historical data set into a first model for training, and determine the preset coefficients of the first model; input the predetermined feature information into a second model for training, and determine the The preset rules of the second model; input the stay duration and the second historical data set into the third model for training, determine the preset threshold of the outlier degree factor, so that the output judgment result of abnormal stay behavior The accuracy and/or recall rate reaches the target value.

The device for training an abnormal staying behavior recognition model for online car-hailing provided by the embodiment of the present invention may be specifically used to execute the method embodiment provided in FIG. 9, and the specific functions are not repeated here.

The device for training an abnormal staying behavior recognition model for online car-hailing provided by an embodiment of the present invention obtains a training data set from a database. The training data includes positive training data belonging to abnormal staying behaviors and negative training data not belonging to abnormal staying behaviors. Data; for each training data, extract the staying position, staying time, driver service status, and predetermined feature information of the training data; obtain the first historical data set and the second historical data set from the database; the first The historical data set includes the historical stay time of at least one driver in the same service state as the training data, and the second historical data set includes the historical stay time of at least one driver in the same time period and in the same area as the training data ; According to the training data set, the first historical data set and the second historical data set corresponding to each training data, the preset model is trained so that the judgment result of the abnormal stay behavior output by the preset model The accuracy and/or recall rate reaches the target value. Through the training of multiple pairs of online car-hailing abnormal staying behavior recognition models, the abnormal staying behavior recognition model of online car-hailing can be obtained, so that massive online car-hailing travel history data and the model can be used, taking into account the length of stay, location, and service. Multi-dimensional feature information such as status can accurately identify whether there are abnormal staying behaviors in the online car-hailing platform in real time, improve the safety perception ability of the online car-hailing platform and the ability to recognize traffic accidents, driver-passenger conflicts, and protect drivers and passengers. Safety.

Fig. 5 is an exemplary flowchart of a method for identifying abnormal stay behavior of a vehicle according to some embodiments of the present application. In some embodiments, the process 500 may be executed by a processing device (for example, the processing device 110 shown in FIG. 1). For example, the process 500 may be stored in a storage device (for example, the storage device 130 or a storage unit of a processing device) in the form of a program or instruction. When the processor 210 or the module shown in FIG. 4 executes the program or instruction, the process may be implemented. 500. In some embodiments, the process 500 may utilize one or more additional operations not described below, and/or not be completed by one or more operations discussed below. In addition, the order of operations shown in FIG. 5 is not restrictive. As shown in FIG. 5, the process 500 may include the following operations.

Step 501: Acquire real-time stay data of the vehicle. In some embodiments, step 510 may be performed by the acquisition module 410.

In some embodiments, the vehicle may be a vehicle that executes a service request. For example, the vehicle may be a small passenger car, a passenger car, a truck, an electric bicycle, an electric motorcycle, etc., which execute a transportation service request including a ride-hailing service request, a navigation service request, a meal delivery service request, a freight service request, etc. The vehicle 150 is shown in FIG. 1. In some embodiments, the service request may be a real-time service request or a reservation service request. You can define "real-time" or "scheduled" by a preset time threshold. For example, if the user (for example, the user of the terminal 120) wants to receive the service immediately or at a time within a preset time threshold from the current time, the service request may be regarded as a real-time service request. If the user wants to receive the service at a time outside the preset time threshold from the current time, the service request can be regarded as a reservation service request. The preset time threshold can be the system default value, or it can be adjusted under different circumstances. For example, the preset time threshold may be 5 minutes, 10 minutes, 20 minutes, and so on. For another example, different preset time thresholds can be set for different cities, different regions, and different time periods.

In some embodiments, the real-time stay data may include the current stay time and characteristic information related to the stay of the vehicle.

In some embodiments, in order to improve the safety of drivers and passengers, the acquisition module 410 may acquire real-time data when the vehicle stops. Specifically, the obtaining module 410 may obtain real-time data from the vehicle 150 or the terminal 140. The real-time data may include, but is not limited to, location information, time information, some status information and interaction information of persons related to the vehicle (for example, drivers and/or passengers). In some embodiments, the processing device 110 (for example, the acquisition module 410) may acquire positioning data of the vehicle, including real-time speed data, and determine whether the vehicle is staying based on the real-time speed data.

When the vehicle stays, the acquisition module 410 may acquire the status of the service provider of the current stay of the vehicle and the characteristic information related to the vehicle stay from the above real-time data. The characteristic information related to the stay of the vehicle may include the current stay position and the behavior of the personnel related to the vehicle. Wherein, the behavior of the personnel related to the vehicle may include the payment behavior and evaluation behavior of the service requester (for example, the passenger requesting a taxi service). The status of the service provider may include being in a service state (for example, in a taxi service, a service provider such as a driver is in a pickup state, an arrival state, or a service provision state) or in a non-service state (for example, in a taxi service, a service provider such as The driver is in the state of taking orders and receiving the vehicle). The characteristic information related to the current vehicle stay includes, but is not limited to, the passenger's payment behavior, the passenger's evaluation of the driver, whether it is in a billing hot spot, and so on.

In other embodiments of this application, in order to improve the safety of online car-hailing drivers and passengers, the server of the online car-hailing platform can collect real-time data when the online car-hailing platform stops. Specifically, it can be obtained from online car-hailing terminals (including Vehicle terminal, driver terminal or passenger terminal) upload some real-time data to the server of the online car-hailing platform. Specifically, the real-time data may include but not limited to location information, time information, some status information and interaction information of the driver and passengers Wait. More specifically, GPS data of online car-hailing can be obtained, including real-time speed data, and it can be determined whether the online car-hailing is staying based on the real-time speed data.

When a certain online car-hailing stops, the current stay location, current stay duration, driver service status, and predetermined feature information can be obtained from the above real-time data. The driver’s service status includes service status (such as pick-up status, arrival status, service status) or non-service status (such as order status, car collection status); predetermined feature information includes, but is not limited to, passenger’s payment behavior and passenger’s Driver’s evaluation, whether it is in a hot spot for billing, etc.

In this application, the real-time data may be the same as or similar to the real-time stay data.

Step 503: Based on the real-time stay data, use the recognition model to determine whether the vehicle is in an abnormal stay state. In some embodiments, step 503 may be performed by the determining module 420.

In some embodiments, the determination module 420 may use the recognition model to process the real-time stay data to realize the real-time recognition of the abnormal stay behavior of the vehicle. In some embodiments, the abnormal stay may refer to an abnormal parking of a vehicle or a person related to the vehicle (for example, a driver and/or a passenger) due to some reasons. For example, parking due to a breakdown of the vehicle. Another example is parking due to a conflict between the driver and the passenger. Another example is parking due to a traffic accident. In some embodiments, the recognition model may be trained through historical data in a database (for example, the storage device 130). The historical data may include historical stay data of multiple vehicles in historical trips, and the recognition results of corresponding stay behaviors. In some embodiments, the recognition model may be any model that can perform outlier detection, for example, a digital outlier model, a Z-score model, an isolated forest model, a DBSCAN model, and so on. The recognition model may be pre-trained and stored in a storage device (for example, the storage device 130 or the storage unit of the processing device). The determining module 420 may communicate with the storage device to obtain the recognition model.

In some embodiments, the recognition model may be a fusion model, which may include a first sub-recognition model, a second sub-recognition model, and a third sub-recognition model. The recognition model can judge whether the stay of the vehicle is an abnormal stay behavior from different dimensions. As an example, the first sub-recognition model may be a distribution-based recognition model, the second sub-recognition model may be a statistics-based sub-recognition model, and the third model may be a density-based sub-recognition model. The recognition model can accurately recognize the stay of the vehicle from the three dimensions of distribution, statistics and density. The input of each sub-recognition model can be different to determine whether the stay of the vehicle is abnormal from a different angle. For other descriptions of determining whether the vehicle is in an abnormally parked state, reference may be made to other parts of this application, for example, Figs. 6-8.

Optionally, the preset model in this embodiment is a fusion model, which may include a first model, a second model, and a third model, and it is gradually determined from different dimensions whether the online car-hailing currently exhibits an abnormal staying behavior.

In some embodiments, the recognition model may be a single model, and one model may be used to implement the functions of the aforementioned fusion model. For example, the determination module 420 may input the real-time stay data of the vehicle into the recognition model to directly determine whether the stay of the vehicle is an abnormal stay. For example, the output result of the recognition model may be 1 or 0. 1 means abnormal stay, and 0 means not abnormal stay.

In some embodiments, different recognition models can be built for different transportation services. For example, one recognition model can be built for ride-hailing services, and another recognition model can be built for freight services. Using different recognition models for different transportation services can improve the accuracy of determining abnormal behavior when vehicles stay in different transportation services.

In some embodiments, the determining module 420 may respectively use the sub-models included in the recognition model to recognize the abnormal stay of the vehicle. As an example, the determining module 420 may determine whether the current stay duration is an abnormal stay duration based on the first sub-recognition model. At the same time, the determining module 420 may determine whether the characteristic information related to the vehicle stay meets the stay rule based on the second sub-recognition model. In addition, the determining module 420 may also determine the abnormal evaluation value related to the current stay time based on the third sub-model and determine whether the abnormal evaluation value is greater than a preset threshold. When the current stay time is the abnormal stay time, the characteristic information related to the vehicle stay does not meet the stay rules, and/or the abnormal evaluation value is greater than the preset threshold, the determining module 420 may determine that the vehicle In an abnormal state. For example, when the current stay time is the abnormal stay time, or the characteristic information related to the vehicle stay does not meet the stay rules, or the abnormal evaluation value is greater than the preset threshold, the determining module 420 may determine The vehicle is in an abnormal parking state. For another example, when the current stay time is the abnormal stay time and the characteristic information related to the vehicle stay does not meet the stay rules, or when the current stay time is the abnormal stay time and the abnormal evaluation value is greater than the When the threshold is preset, or when the characteristic information related to the vehicle stay does not satisfy the stay rule and the abnormality evaluation value is greater than the preset threshold, the determination module 420 may determine that the vehicle is in an abnormal stay state. For another example, as mentioned above, when the current stay time is the abnormal stay duration, the characteristic information related to the vehicle stay does not meet the stay rules and the abnormal evaluation value is greater than the preset threshold, the determining module 420 It may be determined that the vehicle is in an abnormal state.

In some other embodiments of the present application, the first historical data set and the second historical data set may be obtained from a database. The first historical data set includes the historical stay time of at least one driver in the same service state as the real-time data, and the second historical data set includes at least one driver in the same time period and the same area as the real-time data The length of stay in history. The database stores the historical data of all online car-hailing platforms maintained by the online car-hailing platform. In order to identify whether there are abnormal staying behaviors in the current period of online car-hailing, multiple dimensions of historical data can be obtained from the database, so as to identify abnormal stays from multiple dimensions, and improve the accuracy and recall rate. In this embodiment, the first historical data set mainly considers the dimensions of driver service status, while the second historical data set mainly considers the dimensions of time and space. In addition, in this embodiment, the first historical data set and the second historical data set can be obtained from one driver or multiple drivers (such as drivers who meet certain requirements or all drivers) within a certain period of time (for example, within a month) in the database. ) Obtain the corresponding historical stay time within the range of historical data.

In other embodiments of the present application, it may be possible to identify whether the online car-hailing currently exhibits an abnormal staying behavior based on the real-time data, the first historical data set, the second historical data set, and a preset model. According to the preset model and various acquired data, the recognition of the abnormal staying behavior of the online car-hailing can be realized, and the preset model can be trained through the historical data in the database. The preset model can be any model that can perform outlier detection. Optionally, the preset model in this embodiment is a fusion model, which may include a first model, a second model, and a third model, and it is gradually determined from different dimensions whether the online car-hailing currently exhibits an abnormal staying behavior. For details, please refer to Figure 6 of this application.

In this application, the preset model may be the same or similar to the recognition model, the first model may be the same or similar to the first sub-recognition model, and the second model may be the same as the The second sub-recognition model is the same or similar, and the third model may be the same or similar to the third sub-recognition model.

The method for identifying abnormal stays of a vehicle disclosed in this application uses a recognition model trained based on massive vehicle travel history data to process the current stay data of the vehicle, and considers different dimensions such as stay time, location, and status of personnel related to the vehicle This feature information can accurately identify whether the vehicle currently has abnormal staying behavior in real time and accurately, providing the vehicle’s safety perception ability and the ability to recognize traffic accidents and conflicts in the vehicle, such as driver and passenger conflicts, and ensure the safety of personnel.

It should be noted that the foregoing description of the process 500 is only for example and description, and does not limit the scope of application of the present application. For those skilled in the art, various modifications and changes can be made to the process 500 under the guidance of this application. However, these amendments and changes are still within the scope of this application.

Fig. 6 is an exemplary flow chart for determining whether a vehicle is in an abnormal staying state by using a recognition model according to some embodiments of the present application. In some embodiments, the process 600 may be executed by a processing device (for example, the processing device 110 shown in FIG. 1). For example, the process 600 may be stored in a storage device (for example, the storage device 130 or a storage unit of a processing device) in the form of a program or instruction. When the processor 210 or the module shown in FIG. 4 executes the program or instruction, the process may be implemented. 600. In some embodiments, the process 600 may be executed by the determining module 420. In some embodiments, the process 600 may utilize one or more additional operations not described below, and/or not be completed by one or more operations discussed below. In addition, the order of operations shown in FIG. 6 is not restrictive. As shown in FIG. 6, the process 600 may include the following operations.

Step 601: Determine whether the current stay duration is an abnormal stay duration based on the first sub-recognition model.

In some embodiments, the determining module 420 may input the current stay duration into the first sub-recognition model to determine whether the current stay duration is an abnormal stay duration. In some embodiments, the first sub-recognition model may be any recognition model constructed based on an outlier detection algorithm. For example, the outlier detection algorithm may include clustering-based methods, isolated forests, statistical methods, distribution-based Tukey Method and so on. The first sub-recognition model can determine whether the current stay time is an outlier relative to the historical stay time used to train the first sub-recognition model. If yes, it can be determined that the current stay duration is the abnormal stay duration. For other descriptions of determining whether the current stay duration is an abnormal stay duration based on the first sub-recognition model, reference may be made to FIG. 7 of the present application, which will not be repeated here.

Step 603: In response to the current stay duration being the abnormal stay duration, determine whether the characteristic information related to the vehicle stay meets the stay rule based on the second sub-recognition model;

In some embodiments, the second recognition sub-model may be a rule model based on statistics. The determining module 420 may determine whether the characteristic information related to the vehicle stay meets the stay rule determined based on the second recognition sub-model. It is understandable that when it is determined that the current stay duration of the vehicle belongs to the abnormal stay duration, there may also be a long stay caused by some normal reasons. For example, because the vehicle is in a congested road section, the waiting time is too long, so that the first sub-recognition model determines that the current stay time is the abnormal stay time. For another example, after a vehicle that provides a ride-hailing service completes the service, such as arriving at the destination, the passenger pays on the car or gives the driver a good comment. For example, the ride-hailing software installed on the terminal 120 is used to pay or score. The time is too long, resulting in the first child The recognition model determines that the current stay time is the abnormal stay time. Therefore, the determination module 420 can make the stay rule filter the previous determination result. For other descriptions of determining whether the characteristic information related to the vehicle stay meets the stay rule based on the second sub-recognition model, please refer to FIG. 8 of the present application, which will not be repeated here.

In other embodiments of the present application, if it is determined that the current stay duration is the abnormal stay duration, the predetermined characteristic information is input into the second model, and it is determined whether the predetermined characteristic information meets the expectations of the second model. Set rules. The second model is a rule model based on statistics. Some rules can be set in advance to filter some cases where the current stay time is abnormal stay time but not abnormal stay behavior. For example, when staying occurs in some cases, passengers take the initiative to pay and take the initiative Give good comments, or if the current stay is in some billing hotspots, there may also be long stays, but they are not abnormal staying behaviors, that is, the probability of abnormal staying behaviors when these characteristics appear is very low, and the second model can be used on a large scale Reduce the amount of data and reduce the consumption of computing resources. The second preset rule can be determined in advance through training.

In the embodiment, at least one of the predetermined characteristic information in the real-time data, such as the passenger's payment behavior, the passenger's evaluation of the driver, and the billing hotspot area, can be input into the second model to determine whether the passenger's payment behavior is active payment And/or determine whether the passenger’s evaluation of the driver is positive; and/or determine whether the current location is in the billing hotspot area; if the result of at least one of the above determinations is yes, then the reservation is determined The characteristic information satisfies the preset rules of the second model, thereby determining that there is no abnormal stay behavior in the online car-hailing; if the above judgment results are all no, it is determined that the predetermined characteristic information does not meet the preset rules of the second model, The next step can be judged.

In this application, the preset rule may be the same as or similar to the stay rule.

Step 605: In response to the characteristic information related to the vehicle staying that does not satisfy the staying rule, determine the abnormal evaluation value related to the current staying time based on a third sub-recognition model.

In some embodiments, if it is determined that the characteristic information related to the vehicle stay does not meet the stay rules of the second model, the determining module 420 may input the current stay time into the third sub-recognition model to obtain the current stay The abnormal evaluation value of the duration.

In some embodiments, the determining module 420 may obtain the abnormal evaluation value of the current stay time relative to the sample data for training the model through the third sub-recognition model. The sample data for training the model may include second historical stay data of at least one second sample vehicle in the same time period and in the same area. The second historical stay data may include a second historical stay duration. The determining module 420 may use the second sub-recognition model to determine the magnitude of the difference between the current stay duration and the second historical stay duration. The magnitude of the difference may be designated as the abnormality evaluation value. In some embodiments, optionally, in order to increase the data quantity of the second historical stay data and improve the accuracy of calculating the abnormal evaluation value, the area where the current stay position is located adopts the geo5 area (5-digit coding unit network) in the geohash of latitude and longitude. Grid, covering an area of about 24 square kilometers).

In some embodiments, the third sub-recognition model may be any model that can determine outliers, for example, Robust Random Cut Forest, Isolation Forest, Elliptic Envelope, OneClassSVM, Gaussian outlier detection, and so on. In some embodiments, the third sub-recognition model may include a density-based local outlier factor model (Local Outlier Factor, LOF model). The determining module 420 may obtain the local abnormality factor (LOF value) of the current stay time through the LOF model, and designate the LOF value as the abnormality evaluation value. The LOF value can be used to reflect the average of the ratio of the local reachable density of the neighborhood point of the current stay time to the local reachable density of the current stay time. When the LOF value is greater than 1, it means that the local reachable density of the current stay time is smaller than the local reachable density of its neighboring points, and the current stay time is more likely to be an abnormal point. Therefore, by determining the magnitude relationship between the abnormal evaluation value of the current stay time and the preset threshold value, it can be determined whether the vehicle has abnormal stay behavior.

Step 607: Determine whether the abnormality evaluation value is greater than the abnormal staying threshold.

In some embodiments, the abnormal stay threshold may be predetermined. In some embodiments, the abnormal stay threshold may be obtained from a training process. For example, it is determined by pre-determining before training and continuously adjusting during training. The determination module 420 may compare the abnormality evaluation value with the abnormal stay threshold value to determine whether the abnormality evaluation value is greater than the abnormal stay threshold value.

Step 609: In response to the abnormality evaluation value being greater than the abnormal stop threshold, it is determined that the vehicle is in an abnormal stop state.

In some embodiments, when the abnormality evaluation value is greater than the abnormal stay threshold, the determination module 420 may determine that the vehicle is in an abnormal stay state.

In some embodiments, since training the LOF model requires a certain amount of data, that is, the training of the LOF model requires that the number of samples in the second historical stay data is not less than a certain preset number. If the number of samples in the second historical stay data is less than the preset number, the accuracy of the LOF model is relatively poor. At this time, the determining module 420 may directly determine that the vehicle is currently exhibiting abnormal parking behavior when the second sub-recognition model determines that the characteristic information related to the vehicle stay does not meet the staying rules of the second sub-recognition model, and does not need to pass through. The LOF model determines the abnormal evaluation value of the current stay time.

In some embodiments, the preset number can be obtained through the following operations. Count the number of historical stay data (for example, data in the same time period corresponding to the real-time stay data) in the database (for example, the storage device 130 or the memory of the processing device 110) storing the second historical stay data The number of copies or data size), and the result obtained by multiplying the number of historical stay data by a preset percentage (for example, 25%) is used as the preset number.

In this embodiment, the number of stays under different dimensional conditions is obtained, that is, the total number of historical stays (the number of historical stay data) of all vehicles in each geo5 area and the same time period in the statistical database is multiplied by The preset percentage obtains the preset quantity. When the number of samples in the second historical stay data is not less than the preset number, the LOF model is used for the subsequent calculation process. The preset percentage in this embodiment can be obtained through pre-training.

In some other embodiments of the present application, if it is determined that the predetermined characteristic information does not satisfy the preset rules of the second model, the current stay time and the second historical data set are input into the third model , Acquiring the outlier degree factor of the current stay duration, and if the outlier degree factor is greater than a preset threshold, it is determined that the online car-hailing currently exhibits an abnormal stay behavior. In the embodiment, the third model is used to make a final judgment on the current stay time, and the outlier degree factor of the current stay time relative to the second historical data set is obtained to determine that the current stay time is the same time period as each driver in the database. The size of the difference between the historical stay time under different states in the same area. If the outlier degree factor is greater than the preset threshold, the difference is large, so that it can be determined that the current abnormal stay behavior of online car-hailing is present. Optionally, in order to increase the number of data in the second historical data set and improve the accuracy of calculating the outlier degree factor, in this embodiment, the area of the current stay position adopts the geo5 area in the geohash of latitude and longitude (a grid of 5-bit coding units, Covering an area of about 24 square kilometers).

Further, the third model in this embodiment is a density-based LOF (Local Outlier Factor) model; through the LOF model, the current stay duration is obtained according to the current stay duration and the second historical data set The LOF value of the outlier degree factor is used to reflect the average of the ratio of the local reachable density of the neighborhood point of the current stay time to the local reachable density of the current stay time. When the LOF value is greater than 1, it means the current The more the local reachable density of the stay time is smaller than the local reachable density of its neighboring points, the more likely the current stay time is to be an abnormal point. Therefore, it is determined that the LOF value of the outlier degree factor of the current stay time is greater than the preset threshold value, and the network can be determined The ride-hailing is currently exhibiting abnormal stopping behavior. The preset threshold can be obtained through training in advance.

On the basis of the above embodiment, since the LOF model requires a certain amount of data, that is, the number of samples in the second data set is not less than a certain preset number. If the number of samples in the second data set is less than the preset number, then LOF The accuracy of the model is relatively poor. When the second model determines that the predetermined feature information does not meet the preset rules of the second model, it can be directly determined that the online car-hailing is currently exhibiting abnormal staying behavior, without the need to pass The LOF model calculates the LOF value of the outlier degree factor for the current length of stay.

The preset number can be obtained through the following process: Count the number of historical stay data of all drivers in the database during the same time period as the real-time data, and multiply the number of historical stay data by a preset percentage ( For example, the result obtained after 25%) is used as the preset amount.

In this embodiment, the number of stays under different dimensional conditions is obtained, that is, the total number of historical stays (that is, the number of historical stay data) of all drivers in each geo5 area in the same time period in the statistics database, and multiply by The preset percentage obtains the preset quantity. When the number of samples in the second data set is not less than the preset number, the LOF model is used for the subsequent calculation process. The preset percentage in this embodiment can be obtained through pre-training.

On the basis of any of the foregoing embodiments, in the first model, if it is determined that the current stay duration is not an abnormal stay duration, it is determined that the online car-hailing does not currently exhibit abnormal stay behavior; or

In the second model, if it is determined that the predetermined characteristic information satisfies the preset rules of the second model, it is determined that the online car-hailing does not currently exhibit abnormal stay behavior; or

In the third model, if it is determined that the outlier degree factor is less than or equal to a preset threshold, it is determined that the online car-hailing does not currently exhibit abnormal staying behavior.

It should be noted that if it is determined in a certain model that there is no abnormal staying behavior in the online car-hailing currently, there is no need to perform subsequent model processing.

In addition, if it is recognized that the current abnormal staying behavior of online car-hailing is present, human intervention can be carried out, such as the customer service to understand the relevant situation, solve the problem in time, and improve the safety of drivers and passengers.

The method for identifying abnormal staying behaviors of online car-hailing provided by this embodiment collects real-time data when the car-hailing stops. The real-time data includes current staying position, current staying time, driver service status, and predetermined characteristic information; from a database The first historical data set and the second historical data set are obtained in the process; the first historical data set includes the historical stay time of at least one driver in the same service state as the real-time data, and the second historical data set includes at least one The historical stay time of the driver in the same time period and in the same area as the real-time data; according to the real-time data, the first historical data set, the second historical data set, and a preset model, the online appointment is identified Whether the car currently exhibits abnormal stopping behavior. This embodiment uses massive multi-dimensional historical data and preset models for online car-hailing travel, and takes into account the multi-dimensional feature information such as stay time, location, service status, etc., to accurately identify whether there is any abnormal staying behavior in online car-hailing in real time. , Improve the safety perception ability of the online car-hailing platform and the ability to recognize traffic accidents and driver-passenger conflicts, and ensure the safety of drivers and passengers.

In this application, the LOF value of the cluster degree factor may be the same as or similar to the abnormality evaluation value.

In some embodiments, when the determination module 420 is performing the above operations, in the above determination, if the current stay time is not the abnormal stay time, or the characteristic information related to the vehicle stay meets the stay rules, it is determined If the vehicle is not in an abnormal parking state, or the abnormal evaluation value is less than or equal to the preset threshold value, the determination module 420 may determine that the vehicle is not in an abnormal parking state. For example, in step 601, if the current stay time is not the abnormal stay time length, the determination module 420 may directly determine that the vehicle is not in the abnormal stay state. For another example, in step 603, if the characteristic information related to the stay of the vehicle satisfies the stay rule, the determination module 420 may directly determine that the vehicle is not in an abnormal stay state. For another example, if in step 607, if the abnormality evaluation value is less than or equal to the preset threshold value, the determination module 420 may determine that the vehicle is not in an abnormal parking state.

It should be noted that the foregoing description of the process 600 is only for example and description, and does not limit the scope of application of the present application. For those skilled in the art, various modifications and changes can be made to the process 600 under the guidance of this application. However, these amendments and changes are still within the scope of this application.

Fig. 7 is an exemplary flowchart for determining whether the current stay duration is an abnormal stay duration according to some embodiments of the present application. In some embodiments, the process 700 may be executed by a processing device (for example, the processing device 110 shown in FIG. 1). For example, the process 700 may be stored in a storage device (for example, the storage device 130 or the storage unit of the processing device) in the form of a program or instruction. When the processor 210 or the module shown in FIG. 4 executes the program or instruction, the process may be implemented. 700. In some embodiments, the process 700 may be executed by the determining module 420. In some embodiments, the process 700 may utilize one or more additional operations not described below, and/or not be completed by one or more operations discussed below. In addition, the order of operations shown in FIG. 7 is not restrictive.

In some embodiments, the first sub-recognition model may be any recognition model constructed based on an outlier detection algorithm. For example, the outlier detection algorithm may include clustering-based methods, isolated forests, statistical methods, distribution-based Tukey Method and so on. The first sub-recognition model can determine whether the current stay time is an outlier relative to the historical stay time used to train the first sub-recognition model. In some embodiments, the first sub-recognition model includes a Duke's post-comparison model based on distribution. As shown in Figure 7, the specific steps of the method are as follows:

Step 701: Determine the identification coefficient based on the distribution-based Dukai's post-comparison model.

The distribution-based Dukai's post-comparison model can determine the minimum anomaly estimate and the maximum anomaly estimate in the data set for training the model based on the quartile of the data set for training the model and the identification coefficient k used to characterize the degree of anomaly . Values outside the range of the minimum anomaly estimated value and the maximum anomaly estimated value may be an outlier. The greater the identification coefficient k used to characterize the degree of abnormality, the greater the degree of abnormality of the value beyond the minimum and maximum estimated value of abnormality. For example, when k=1.5, the above-mentioned abnormality is moderately abnormal, and when k=3 The above-mentioned degree of abnormality is extremely abnormal.

In some embodiments, the identification coefficient k may be predetermined and determined, for example, k=1.5, k=2, k=2.5, k=3, and so on. In some embodiments, the recognition coefficient k may be determined through the training process of the first sub-recognition model. For example, it is determined by pre-determining before training and continuously adjusting during training. This application does not specifically limit the determination method of the identification coefficient.

Step 703: Determine at least one maximum abnormality estimated value based on the identification coefficient.

In some embodiments, the Dukai method can obtain the maximum abnormality estimate of the sample data used to train the model by comparing the model after the fact. The sample data of the training Dukai method post-comparison model may include first historical stay data of at least one first sample vehicle in the same vehicle state. The same vehicle state may mean that the vehicles are in the same state, or that the service providers are in the same state. For example, assuming that the service provider (for example, the driver) corresponding to the vehicle is in the service state (for example, the driving state), the service provider corresponding to the first sample vehicle in the sample data of the training Du Kaifa post-comparison model (For example, the driver) is also in the service state (for example, the driving state). The first historical stay data may at least include historical stay time. If the current stay duration is greater than the maximum abnormality estimated value, it is determined that the current stay duration is the abnormal stay duration. As an example, the determining module 420 may sort the historical stay time from smallest to largest, and then take the data located at the 25th percentile in the ranking as Q1, the data at the 50th percentile as Q2, and the data at the 75th percentile as Q3. , And then determine the maximum abnormality estimate of the first historical stay data=Q3+k(Q3-Q1).

In some embodiments, the sample data for training the Dukai method post-comparison model may include a first subset and a second subset. The first subset may include the first historical stay duration in the same state of the service provider (for example, driver) corresponding to the vehicle and the service provider included in the real-time stay data. The second subset includes that all service providers (for example, drivers) in the sample data for training the Du Kaifa post-comparison model are in the same state as the status of the service providers (for example, drivers) included in the real-time stay data, and The second historical stay in the same area. In some embodiments, through the first subset, it can be determined whether the current stay time is an abnormal value relative to the historical stay time of the service provider (for example, the driver) corresponding to the vehicle in the same state. The personal habits of the service provider (for example, the driver) corresponding to the vehicle in the same state. Through the second subset, it can be determined whether the current stay time is an abnormal value relative to the historical stay time of all service providers (for example, drivers) in the same service state and in the same area, and the service provision corresponding to different vehicles can be considered. The general condition of a person (for example, a driver) staying in the same service state and in the same area. This application considers the service status dimension of the service provider (for example, the driver) corresponding to the vehicle, and takes into account the personal habits of the service provider (for example, the driver) corresponding to the vehicle and the status of the area where the current location belongs. It can be more accurately determined whether the current stay duration is abnormal stay duration. In some embodiments, the area in the above description adopts the geo6 area in the geohash of latitude and longitude (a 6-bit coding unit grid, covering an area of about 0.73 square kilometers).

Similarly, based on the identification coefficient, the determining module 420 may obtain the first maximum abnormality estimation value of the first subset and the second maximum abnormality estimation value of the second subset.

In some embodiments, there may be different maximum abnormality estimates for different states of service providers (for example, drivers) corresponding to the vehicles. For example, for the service provider (for example, the driver) corresponding to the vehicle in the service state and the non-service state, the maximum abnormality estimated value may correspond to it respectively.

Step 705: Determine whether the current stay time is greater than at least one of the at least one maximum abnormality estimation value.

In some embodiments, the at least one maximum abnormality estimate may include the maximum abnormality estimate of the first historical stay data, the first maximum abnormality estimate of the first subset, and the second maximum abnormality estimate of the second subset value. The determining module 420 may compare the current stay time with each of the at least one maximum abnormality estimated value to determine the magnitude between the current stay time and each of the at least one maximum abnormality estimated value relationship.

Step 707: In response to the current stay time being greater than at least one of the at least one maximum abnormality estimation value, it is determined that the current stay time length is the abnormal stay time length.

In some embodiments, if the current stay duration is greater than at least one of the at least one maximum abnormality estimation value, the determining module 420 may determine that the current stay duration is the abnormal stay duration.

In other embodiments of the present application, the current stay time length and the first historical data set may be input into a first model to determine whether the current stay time length is abnormal relative to the first historical data set Length of stay. The first model may preliminarily determine whether the current stay time is an abnormal stay time based on the first historical data set, that is, determine whether the current stay time is a possible outlier relative to the first historical data set, where the first model may be Existing arbitrary outlier detection methods, such as clustering-based methods, isolated forests, statistical methods, and distribution-based Tukey Method methods.

In an optional embodiment, the first model is a distribution-based Tukey Method model, which usually can be based on the quartile of the data set and the coefficient k used to characterize the degree of abnormality, and the minimum estimate in the data set can be calculated Value and maximum estimated value, the value beyond the minimum estimated value and maximum estimated value range may be an outlier. The larger the coefficient k used to characterize the degree of abnormality, the greater the abnormal degree of the value beyond the range of the minimum estimated value and the maximum estimated value. For example, when k=1.5, the above-mentioned abnormality degree is moderately abnormal, and when k=3, the above-mentioned abnormality degree is extremely abnormal. The Tukey Method model in this embodiment can only obtain the maximum estimated value of the first historical data set, and if the current stay duration is greater than the maximum estimated value, it is determined that the current stay duration is the abnormal stay duration. Specifically, the data in the first historical data set can be sorted from small to large, and then the data located at the 25th percentile in the sorting is recorded as Q1, the data located at the 50th percentile is recorded as Q2, and the data located at the 75th percentile is recorded as Q1. Mark it as Q3, and then calculate the maximum estimated value of the first historical data set=Q3+k(Q3-Q1). The coefficient k used to characterize the degree of abnormality adopts a preset coefficient, and the specific value can be determined in advance through the training process.

Further, the first historical data set includes a first subset and a second subset; the first subset includes the historical stay time of the current driver in the same service state as the real-time data; the second subset Including the historical stay time of all drivers in the database in the same service state and the same area as the real-time data; further, the judging whether the current stay time is abnormal stay relative to the first historical data set Duration, including:

In the embodiment, through the first subset, it can be judged whether the current stay time is an abnormal value relative to the historical stay time of the current driver in the same state, and the personal habits of the current driver in the same state can be considered; through the second sub-set Set, it can be judged whether the current stay time is an abnormal value relative to the historical stay time of all drivers in the same service state and the same area. It can take into account the normal situation of different drivers staying in the same service state and the same area. In this embodiment, while considering the dimension of the driver's service status, it also takes into account the current driver's personal habits and the conditions of the area to which the current stay location belongs, and can more accurately determine whether the current stay duration is an abnormal stay duration. Optionally, the area to which the current staying location belongs in this embodiment adopts the geo6 area in the geohash of latitude and longitude (a 6-bit coding unit grid, covering an area of about 0.73 square kilometers).

It should be noted that the foregoing description of the process 700 is only for example and description, and does not limit the scope of application of the present application. For those skilled in the art, various modifications and changes can be made to the process 700 under the guidance of this application. However, these amendments and changes are still within the scope of this application.

Fig. 8 is an exemplary flowchart for determining whether the characteristic information related to the current vehicle stay meets the stay rule according to some embodiments of the present application. In some embodiments, the process 800 may be executed by a processing device (for example, the processing device 110 shown in FIG. 1). For example, the process 800 may be stored in a storage device (for example, the storage device 130 or the storage unit of the processing device) in the form of a program or instruction. When the processor 210 or the module shown in FIG. 4 executes the program or instruction, the process may be implemented. 800. In some embodiments, the process 800 may be executed by the determining module 420. In some embodiments, the process 800 may utilize one or more additional operations not described below, and/or not be completed by one or more operations discussed below. In addition, the order of operations shown in FIG. 8 is not restrictive.

Step 801: Determine the stay rule based on the second sub-recognition model.

In some embodiments, the second sub-recognition model may include a rule model based on statistics. It can be known that although the current stay duration is determined to be the abnormal stay duration, there are also cases where the stay does not belong to the abnormal stay behavior. For example, after the vehicle completes the service, for example, after arriving at the destination, the service requester, such as the passenger, pays in the vehicle or gives the driver a favorable comment, such as using the taxi software installed on the terminal 120 to pay or score, or the service requester, such as the passenger, has special requirements/ Demand (for example, waiting for children to leave school), or service providers such as drivers actively praise service requesters such as passengers, or the current location of the vehicle is some service request hot spot area or traffic jam area. That is to say, when the above situation occurs, even if the current stay duration is determined as the abnormal stay duration, the probability of the abnormal stay behavior of the vehicle is very low. Therefore, the determination module 420 makes a determination through the second sub-recognition model to reduce subsequent consumption of computing resources. In some embodiments, the stay rule may be determined based on training preset by the second sub-recognition model. For example, by statistically analyzing the historical stay data of multiple vehicles to obtain stay rules.

Step 803: Determine whether the characteristic information related to the stay of the vehicle satisfies the stay rule.

In some embodiments, the characteristic information related to the stay of the vehicle may include at least the behavior of the personnel related to the vehicle and the current stay position of the vehicle. The behavior of the personnel related to the vehicle may include the behavior of the service requester and/or the service provider. The behavior of the service requester may include payment behavior, evaluation behavior, conflict behavior, etc. of the service requester (for example, a passenger requesting a taxi service). The behavior of the service provider may include the evaluation behavior, service behavior, conflict behavior, etc. of the service provider (for example, a driver who provides a taxi service). The determining module 420 may determine whether the behavior of the person related to the vehicle is a normal behavior, and/or determine whether the current stay location is in a hot spot area. The normal behavior may include payment behavior (for example, active payment), evaluation behavior (for example, active praise), and the like. Abnormal behavior may include conflict behavior (for example, driver and passenger conflict) and the like. The hot spot area may include a service request sending hot spot area and/or a traffic jam area. If the determination module 420 determines that the determination result of at least one of the above items is yes, it can be determined that the characteristic information related to the current vehicle stay meets the stay rules preset by the second model, so as to determine that the vehicle currently does not exhibit abnormal stay behavior. If the above judgment results are all no, it is determined that the characteristic information related to the current vehicle stay does not meet the stay rules preset by the second model, and the third sub-recognition model can make the next determination.

It should be noted that the foregoing description of the process 800 is only for example and description, and does not limit the scope of application of the present application. For those skilled in the art, various modifications and changes can be made to the process 800 under the guidance of this application. However, these amendments and changes are still within the scope of this application.

Fig. 9 is an exemplary flowchart of a method for training a recognition model according to some embodiments of the present application. In some embodiments, the process 900 may be executed by a processing device (for example, the processing device 110 shown in FIG. 1). For example, the process 900 may be stored in a storage device (for example, the storage device 130 or a storage unit of a processing device) in the form of a program or instruction. When the processor 210 or the module shown in FIG. 4 executes the program or instruction, the process may be implemented. 900. In some embodiments, the process 900 may be executed by the training module 430. In some embodiments, the process 900 may utilize one or more additional operations not described below, and/or not be completed by one or more operations discussed below. In addition, the order of operations shown in FIG. 9 is not restrictive.

Step 901: Obtain multiple training samples.

The multiple training samples include first historical stay data of at least one first sample vehicle in the same vehicle state, and second historical stay data of at least one second sample vehicle in the same time period and in the same area.

In some embodiments, the multiple training samples include positive samples belonging to abnormal staying behaviors and negative samples not belonging to abnormal staying behaviors. Among them, the positive samples may include data when the vehicle has abnormal stopping behaviors in different scenarios. For example, abnormal stop only occurs in the service state (for example, the driver conflicts with the passenger after the passenger gets on the bus or the passenger is about to get off in the taxi service, and the service order is not ended), and the abnormal stop only occurs in the non-service state (for example, In the taxi service, the driver ends the service order during the driving process, stays in the non-service state, and conflicts with the passenger), and stops abnormally in the service state and the non-service state (for example, the driver stays in the service in the taxi service and ends the order , And began to clash with passengers). Conflicts with passengers can include: vehicle-related situations (such as passengers damaging the vehicle, dirtying the vehicle), cost-related situations (such as passengers disagreeing with service fees), and driver safety-related situations (such as passenger abuse or assault) Drivers), situations related to traffic accidents (such as sudden brakes causing injury to passengers, or other traffic accidents causing injury to passengers). Negative samples can include data where the vehicle has stayed for a long time but does not belong to abnormal parking behavior, and data where the vehicle has not stayed for a long time. Among them, the positive sample and the negative sample specifically include vehicle-related location information, time information, status information and interaction information of personnel related to the vehicle, and so on.

In some embodiments, each training sample includes the historical stay time of the vehicle, the status of the service provider (such as driving status, arrival status, service status) or non-service status (such as order status, receiving status), and current Feature information related to vehicle stays. The characteristic information related to the stay of the vehicle may include the current stay position and the behavior of the personnel related to the vehicle. Wherein, the behavior of the personnel related to the vehicle may include the payment behavior and evaluation behavior of the service requester (for example, the passenger requesting a taxi service).

In some embodiments, the multiple training samples include the first historical stay data of at least one first sample vehicle in the same vehicle state, and the second historical stay data of at least one second sample vehicle in the same time period and in the same area. Historical stay data. The multiple training samples include first historical stay data of at least one first sample vehicle in the same vehicle state, and second historical stay data of at least one second sample vehicle in the same time period and in the same area.

In some embodiments, the database (for example, the storage device 130 or the storage unit of the processing device) stores historical data of all vehicles that have performed transportation services. In order to identify whether the vehicle has abnormal staying behavior, multiple dimensions of historical data can be obtained from the database, so as to identify abnormal stays from multiple dimensions, and improve the accuracy and recall rate. The first historical stay data mainly considers the dimensions of the state of the service provider corresponding to the vehicle, while the second historical stay data mainly considers the dimensions of time and space. In addition, the first historical stay data and the second historical stay data may be from the historical data of one or more service providers (such as drivers who meet certain requirements or all drivers) within a certain period of time (for example, within a month) in the database Obtain the corresponding historical stay data within the range of.

In some embodiments of the present application, a training data set may be obtained from a database, and the training data includes positive training data belonging to abnormal staying behaviors and negative training data not belonging to abnormal staying behaviors. In order to train the above-mentioned preset model, a training data set can be obtained from the database, where the positive training data can include data on abnormal staying behaviors of online ride-hailing in different scenarios, for example, abnormal staying only in the service state ( The driver conflicted with the passenger after the passenger got on the bus or the passenger was about to get off the bus, and did not end the order), and only stopped abnormally in the non-service state (the driver ended the order while driving, stayed in the non-service state, and conflicted with the passenger ), abnormal stays occur in service status and non-service status (the driver stays in service, ends the order, and begins to conflict with passengers). Specifically, conflicts with passengers can include: vehicle-related situations (such as passenger damage Vehicles, dirty vehicles), expenses-related circumstances (such as passengers disagreeing with the fees), driver safety-related circumstances (such as passengers verbally or assaulting the driver), traffic accident-related circumstances (such as sudden braking and injury to passengers, Or other traffic accidents caused passengers to be injured). Negative training data may include data that has been staying for a long time in online car-hailing but does not belong to abnormal staying behavior, and data that has not stayed for a long time in online car-hailing. Among them, the positive training data and negative training data specifically include location information, time information, some status information and interaction information of the driver and passengers related to the online car-hailing.

In some embodiments of the present application, for each training data, the stay location, stay duration, driver service status, and predetermined feature information of the training data can be extracted. For each of the above training data, extract the training data's staying location, staying time, driver service status (such as driving status, arrival status, service status) or non-service status (such as order status, receiving status) and predetermined characteristics Information (including but not limited to at least one of the passenger's payment behavior, the passenger's evaluation of the driver, and whether it is in the billing hotspot area). For the specific process, please refer to the process of 501 in FIG. 5 above.

In some embodiments of the present application, the first historical data set and the second historical data set may be obtained from the database; the first historical data set includes at least one driver in the same service state as the training data Historical stay time, the second historical data set includes the historical stay time of at least one driver in the same time period and in the same area as the training data. For each training data, the first historical data set and the second historical data set are obtained. For obtaining the first historical data set and the second historical data set, please refer to the process of 503 in FIG. 5, which will not be repeated here.

Step 903: Train a first initial sub-recognition model based on the first historical stay data to obtain a first sub-recognition model and recognition parameters related to the first sub-recognition model.

In some embodiments, the first initial sub-recognition model may be any recognition model constructed based on an outlier detection algorithm. For example, the outlier detection algorithm may include clustering-based methods, isolated forests, statistical methods, distribution-based Tukey Method and so on. The training module 430 may use the first historical stay data to train the first sub-recognition model while determining the recognition parameters. As an example, assuming that the first sub-recognition model is a distribution-based Dukai's post-comparison model, the training module 430 can set an initial recognition parameter before training, and compare the prediction results for each training sample during the training process. (For example, whether it is an abnormal stay) and the actual situation when the vehicle corresponding to the training sample stays, adjust the initial recognition parameters and/or the model parameters of the first initial sub-recognition model. Until the model is trained.

Step 905: Train a second initial sub-recognition model based on the first historical stay data and the second historical stay data to obtain a second sub-recognition model and stay rules related to the second sub-recognition model.

In some embodiments, the second initial sub-recognition model may include a rule model based on statistics. The training module 430 may perform statistical analysis on the training samples to construct the second initial sub-model. For example, the training samples whose historical stay duration included in the sample belong to the abnormal stay duration but are negative samples are analyzed to determine one or more preset rules. Subsequently, the final stay rule can be obtained through continuous debugging and/or modification of the preset rule. At the same time, the trained second sub-recognition model can be obtained.

Step 907: Train a third initial sub-recognition model based on the second historical stay data to obtain a third sub-recognition model and an abnormal stay threshold related to the third sub-recognition model.

In some embodiments, the third initial sub-recognition model can be any model that can determine outliers, for example, Robust Random Cut Forest, Isolation Forest, Elliptic Envelope, OneClassSVM, Gaussian outlier detection, and so on. In some embodiments, the third sub-recognition model may include a density-based local outlier factor model (Local Outlier Factor, LOF model). The training module 430 may use the second historical stay data to train the third initial sub-recognition model, for example, continuously adjust the abnormal stay threshold used to evaluate the local abnormality factor (LOF value) to obtain the third sub-recognition model, and the final Abnormal stay threshold.

The execution subject of the vehicle's abnormal stay behavior recognition model training method can be the same as the execution subject of the vehicle's abnormal stay behavior recognition method, of course, it can also be different.

In some embodiments of the present application, a preset model may be trained based on the training data set, the first historical data set and the second historical data set corresponding to each training data, so that the preset model The accuracy and/or recall rate of the abnormal stay behavior judgment result output by the model reaches the target value.

In the embodiment, according to each training data in the training data set, and the first historical data set and the second historical data set corresponding to the training data, the preset model is trained. During the training process, each training data , And the first historical data set and the second historical data set corresponding to the training data are input into the preset model, and the process is the same as that in Figure 6 to Figure 8, and the preset model output is realized by debugging the relevant parameters of the preset model The accuracy and/or recall rate of the judgment result of abnormal stay behavior reached the target value.

More specifically, the preset model is a fusion model, which may include a first model, a second model, and a third model. It is gradually determined from different dimensions whether the online car-hailing is currently exhibiting abnormal staying behavior. The step S304 is based on the The training data set, the first historical data set and the second historical data set corresponding to each training data, and the training of the preset model may include:

For each training data, input the stay duration and the first historical data set into a first model for training, and determine the preset coefficients of the first model;

Inputting the predetermined feature information into a second model for training, and determining a preset rule of the second model;

The staying time and the second historical data set are input into a third model for training, and a preset threshold value of the outlier degree factor is determined.

In this embodiment, the training process for the first model, the second model, and the third model is similar to the process of Fig. 6 and Fig. 7, and the abnormal stay behavior of the output is realized by repeatedly debugging the relevant parameters in each model. The accuracy and/or recall rate of the judgment result reaches the target value. For example, if the first model is a distribution-based Tukey Method model, the coefficients used to characterize the degree of abnormality need to be debugged multiple times; if the second model is a statistically-based rule model, some rules can be set in advance, and then the relevant thresholds of each rule can be debugged ; If the third model is a density-based LOF model, the preset threshold for evaluating the LOF value of the outlier degree factor needs to be debugged multiple times; if the second data set has a large number of samples, the LOF model is not performed The calculation directly determines that the online car-hailing is currently abnormally staying, and the preset number of related parameters also need to be debugged multiple times.

The execution subject of the method for identifying abnormal staying behavior recognition model for online car-hailing in this embodiment may be the same as the execution subject of the method for identifying abnormal staying behavior for online car-hailing, of course, it may also be different.

The method for training an abnormal staying behavior recognition model for online car-hailing provided in this embodiment is to obtain a training data set from a database, and the training data includes positive training data belonging to abnormal staying behaviors and negative training data not belonging to abnormal staying behaviors For each training data, extract the staying position, staying time, driver service status and predetermined feature information of the training data; obtain the first historical data set and the second historical data set from the database; the first history The data set includes the historical stay time of at least one driver in the same service state as the training data, and the second historical data set includes the historical stay time of at least one driver in the same time period and in the same area as the training data; According to the training data set, the first historical data set and the second historical data set corresponding to each training data, the preset model is trained to make the judgment result of the abnormal staying behavior output by the preset model accurate The rate and/or recall rate reach the target value. Through the training of multiple pairs of online car-hailing abnormal staying behavior recognition models, the abnormal staying behavior recognition model of online car-hailing can be obtained, so that massive online car-hailing travel history data and the model can be used, taking into account the length of stay, location, and service. Multi-dimensional feature information such as status can accurately identify whether there are abnormal staying behaviors in the online car-hailing platform in real time, improve the safety perception ability of the online car-hailing platform and the ability to recognize traffic accidents, driver-passenger conflicts, and protect drivers and passengers. Safety.

It should be noted that the foregoing description of the process 900 is only for example and description, and does not limit the scope of application of the present application. For those skilled in the art, various modifications and changes can be made to the process 900 under the guidance of this application. However, these amendments and changes are still within the scope of this application.

The embodiment of the present application also provides a schematic structural diagram of a device for identifying abnormal stay behavior of online car-hailing. The device for identifying abnormal staying behaviors of online car-hailing provided in the embodiments of the present application can execute the processing flow provided in the embodiment of the method for identifying abnormal staying behaviors of online car-hailing. The device for identifying abnormal staying behavior of online car-hailing includes a memory, a processor, a computer program, and a communication interface; wherein the computer program is stored in the memory and is configured to execute the online car-hailing operation described in the above embodiment by the processor. Method for identifying abnormal stay behavior.

The device for identifying abnormal stay behavior of online car-hailing in the foregoing embodiment can be used to implement the technical solution of the foregoing method embodiment, and its implementation principles and technical effects are similar, and will not be repeated here.

The embodiment of the present application also provides a schematic structural diagram of a training device for an abnormal stay behavior recognition model for online car-hailing. The device for training an abnormal staying behavior recognition model for online car-hailing provided by the embodiment of the present invention can execute the processing flow provided in the embodiment of the method for training an abnormal staying behavior recognition model for online car-hailing. The training equipment for the recognition model of abnormal staying behavior of online car-hailing includes a memory, a processor, a computer program, and a communication interface; wherein the computer program is stored in the memory and is configured to execute the online car-hailing system described in the above embodiments by the processor. Training method of abnormal stay behavior recognition model.

The training device for the recognition model of abnormal stay behavior of online car-hailing in the foregoing embodiment can be used to implement the technical solutions of the foregoing method embodiments, and its implementation principles and technical effects are similar, and will not be repeated here.

In addition, this embodiment also provides a computer-readable storage medium on which a computer program is stored, and the computer program is executed by a processor to implement the method for identifying abnormal stay behavior of online car-hailing described in the foregoing embodiment.

In addition, this embodiment also provides a computer-readable storage medium on which a computer program is stored, and the computer program is executed by a processor to implement the method for training an abnormal stay behavior recognition model for online car-hailing described in the foregoing embodiment.

The basic concepts have been described above. Obviously, for those skilled in the art, the above detailed disclosure is only an example, and does not constitute a limitation to the application. Although it is not explicitly stated here, those skilled in the art may make various modifications, improvements and amendments to this application. Such modifications, improvements, and corrections are suggested in this application, so such modifications, improvements, and corrections still belong to the spirit and scope of the exemplary embodiments of this application.

At the same time, this application uses specific words to describe the embodiments of the application. For example, "one embodiment", "an embodiment", and/or "some embodiments" mean a certain feature, structure, or characteristic related to at least one embodiment of the present application. Therefore, it should be emphasized and noted that “one embodiment” or “one embodiment” or “an alternative embodiment” mentioned twice or more in different positions in this specification does not necessarily refer to the same embodiment. . In addition, some features, structures, or characteristics in one or more embodiments of the present application can be appropriately combined.

In addition, those skilled in the art can understand that various aspects of this application can be explained and described through a number of patentable categories or situations, including any new and useful process, machine, product, or combination of substances, or a combination of them. Any new and useful improvements. Correspondingly, various aspects of the present application can be completely executed by hardware, can be completely executed by software (including firmware, resident software, microcode, etc.), or can be executed by a combination of hardware and software. The above hardware or software can all be referred to as "data block", "module", "engine", "unit", "component" or "system". In addition, various aspects of the present application may be embodied as a computer product located in one or more computer-readable media, and the product includes computer-readable program codes.

The computer storage medium may contain a propagated data signal containing a computer program code, for example on a baseband or as part of a carrier wave. The propagated signal may have multiple manifestations, including electromagnetic forms, optical forms, etc., or a suitable combination. The computer storage medium may be any computer readable medium other than the computer readable storage medium, and the medium may be connected to an instruction execution system, device, or device to realize communication, propagation, or transmission of the program for use. The program code located on the computer storage medium can be transmitted through any suitable medium, including radio, cable, fiber optic cable, RF, or similar medium, or any combination of the above medium.

The computer program codes required for the operation of each part of this application can be written in any one or more programming languages, including object-oriented programming languages such as Java, Scala, Smalltalk, Eiffel, JADE, Emerald, C++, C#, VB.NET, Python Etc., conventional programming languages such as C language, Visual Basic, Fortran 2003, Perl, COBOL 2002, PHP, ABAP, dynamic programming languages such as Python, Ruby and Groovy, or other programming languages. The program code can be run entirely on the user's computer, or run as an independent software package on the user's computer, or partly run on the user's computer and partly run on a remote computer, or run entirely on the remote computer or server. In the latter case, the remote computer can be connected to the user's computer through any form of network, such as a local area network (LAN) or a wide area network (WAN), or connected to an external computer (for example, via the Internet), or in a cloud computing environment, or as a service Use software as a service (SaaS).

In addition, unless explicitly stated in the claims, the order of processing elements and sequences, the use of numbers and letters, or the use of other names in this application are not used to limit the order of the procedures and methods of this application. Although the foregoing disclosure uses various examples to discuss some embodiments of the invention that are currently considered useful, it should be understood that such details are only for illustrative purposes, and the appended claims are not limited to the disclosed embodiments. On the contrary, the rights are The requirements are intended to cover all modifications and equivalent combinations that conform to the essence and scope of the embodiments of the present application. For example, although the system components described above can be implemented by hardware devices, they can also be implemented only by software solutions, such as installing the described system on an existing server or mobile device.

For the same reason, it should be noted that, in order to simplify the expressions disclosed in this application and thus help the understanding of one or more embodiments of the invention, in the foregoing description of the embodiments of this application, multiple features are sometimes combined into one embodiment. In the drawings or its description. However, this method of disclosure does not mean that the subject of the application requires more features than those mentioned in the claims. In fact, the features of the embodiment are less than all the features of the single embodiment disclosed above.

In some embodiments, numbers describing the number of ingredients and attributes are used. It should be understood that such numbers used in the description of the embodiments use the modifier "about", "approximately" or "substantially" in some examples. Retouch. Unless otherwise stated, "approximately", "approximately" or "substantially" indicates that the number is allowed to vary by ±20%. Correspondingly, in some embodiments, the numerical parameters used in the specification and claims are approximate values, and the approximate values can be changed according to the required characteristics of individual embodiments. In some embodiments, the numerical parameter should consider the prescribed effective digits and adopt the method of general digit retention. Although the numerical ranges and parameters used to confirm the breadth of the range in some embodiments of the present application are approximate values, in specific embodiments, the setting of such numerical values is as accurate as possible within the feasible range.

For each patent, patent application, patent application publication and other materials cited in this application, such as articles, books, specifications, publications, documents, etc., the entire contents of which are hereby incorporated into this application by reference. The application history documents that are inconsistent or conflicting with the content of this application are excluded, and documents that restrict the broadest scope of the claims of this application (currently or later attached to this application) are also excluded. It should be noted that if there is any inconsistency or conflict between the description, definition, and/or term usage in the attached materials of this application and the content described in this application, the description, definition and/or term usage of this application shall prevail .

Finally, it should be understood that the embodiments described in this application are only used to illustrate the principles of the embodiments of this application. Other variations may also fall within the scope of this application. Therefore, as an example and not a limitation, the alternative configuration of the embodiment of the present application can be regarded as consistent with the teaching of the present application. Correspondingly, the embodiments of the present application are not limited to the embodiments explicitly introduced and described in the present application.

Claims

A system including:

At least one storage medium, the storage medium including an instruction for identifying the abnormal stay of the vehicle;

At least one processor, the at least one processor is in communication with the at least one storage medium, wherein, when the instruction is executed, the at least one processor is configured to:

Acquiring real-time stay data of the vehicle;

Based on the real-time stay data, a recognition model is used to determine whether the vehicle is in an abnormal stay state; wherein the recognition model includes a machine learning model trained based on historical stay data of multiple historical vehicles.
The system according to claim 1, wherein the real-time stay data includes the current stay time and characteristic information related to the vehicle stay. Based on the real-time stay data, a trained recognition model is used to determine whether the vehicle is in an abnormal stay. State, the at least one processor is configured to:

Determining whether the current stay time is an abnormal stay time based on the first sub-recognition model;

In response to the current stay time being the abnormal stay time length, determining whether the characteristic information related to the vehicle stay meets the stay rule based on the second sub-recognition model;

In response to the characteristic information related to the vehicle stay that does not satisfy the stay rule, determining the abnormal evaluation value related to the current stay time based on the third sub-model;

Determine whether the abnormality evaluation value is greater than the abnormal stay threshold;

In response to the abnormality evaluation value being greater than the preset threshold value, it is determined that the vehicle is in an abnormal stay state.
3. The system according to claim 2, wherein the first sub-recognition model comprises a distribution-based Dukai's post-comparison model to determine whether the current stay duration is an abnormal stay duration based on the first sub-recognition model. At least one processor is configured as:

Determine the identification coefficient based on the distribution-based Dukai's post-comparison model;

Determining at least one maximum anomaly estimated value based on the identification coefficient;

Determining whether the current stay time is greater than at least one of the at least one maximum abnormality estimation value;

In response to the current stay time being greater than at least one of the at least one maximum abnormality estimation value, it is determined that the current stay time length is the abnormal stay time length.
The system according to claim 2, wherein the characteristic information related to the stay of the vehicle includes at least one of the following: the behavior of the personnel related to the vehicle and the current staying position of the vehicle; it is determined based on the second sub-recognition model Whether the characteristic information related to the stay of the vehicle satisfies the stay rule, the at least one processor is configured to:

Determining the stay rule based on the second sub-recognition model;

Determining whether the characteristic information related to the vehicle stay meets the stay rule includes:

Determine whether the behavior of the person related to the vehicle is a normal behavior; and/or

Determine whether the current stay position is in a hot spot area;

In response to that the behavior of the person related to the vehicle is not a normal behavior, and the current stay position is not in a hot spot area, it is determined that the characteristic information related to the stay of the vehicle does not satisfy the stay rule.
The system according to claim 4, wherein the third sub-recognition model comprises a density-based local abnormality factor algorithm model; to determine the abnormality evaluation value related to the current stay time based on the third sub-model, the at least one The processor is configured as:

The local abnormality factor related to the current stay time is determined based on the local abnormality factor algorithm model, and the local abnormality factor is designated as the abnormality evaluation value.
The system of claim 2, wherein the at least one processor is further configured to:

In response to that the current stay time is not an abnormal stay time, it is determined that the vehicle is not in an abnormal stay state; or

In response to the feature information related to the stay of the vehicle satisfying the stay rule, it is determined that the vehicle is not in an abnormal stay state; or

In response to the abnormality evaluation value being less than or equal to the preset threshold value, it is determined that the vehicle is not in an abnormal stay state.
The system according to claim 1, wherein the real-time stay data includes the current stay time and characteristic information related to the vehicle stay. Based on the real-time stay data, a trained recognition model is used to determine whether the vehicle is in an abnormal stay. State, the at least one processor is configured to:

Determining whether the current stay time is an abnormal stay time based on the first sub-recognition model;

Determining whether the characteristic information related to the vehicle stay meets the stay rule based on the second sub-recognition model;

Determining the abnormal evaluation value related to the current stay time based on the third sub-model and determining whether the abnormal evaluation value is greater than a preset threshold;

In response to the current stay time being the abnormal stay time, the characteristic information related to the vehicle stay does not meet the stay rules, and/or the abnormality evaluation value is greater than the preset threshold, it is determined that the vehicle is in an abnormal stay state .
The system of claim 1, wherein the at least one processor is configured to:

Acquire multiple training samples, the multiple training samples including first historical stay data of at least one first sample vehicle in the same vehicle state, and at least one second sample vehicle in the same time period and in the same area Second historical stay data; the first historical stay data and the second historical stay include historical stay duration and characteristic information related to historical vehicle stays,

Training a first initial sub-recognition model based on the first historical stay data to obtain a first sub-recognition model and recognition parameters related to the first sub-recognition model;

Training a second initial sub-recognition model based on the first historical stay data and the second historical stay data to obtain a second sub-recognition model and stay rules related to the second sub-recognition model;

Training a third initial sub-recognition model based on the second historical stay data to obtain a third sub-recognition model and an abnormal stay threshold related to the third sub-recognition model.
The system of claim 1, wherein the at least one processor is configured to:

Based on the real-time stay data, the recognition model is updated.
An abnormal stay method for identifying vehicles, wherein the method includes:

Acquiring real-time stay data of the vehicle;

Based on the real-time stay data, a recognition model is used to determine whether the vehicle is in an abnormal stay state; wherein the recognition model includes a machine learning model trained based on historical stay data of multiple historical vehicles.
The method according to claim 10, wherein the real-time stay data includes current stay time and characteristic information related to vehicle stay, and based on the real-time stay data, a trained recognition model is used to determine whether the vehicle is abnormal Stay status, including:

Determining whether the current stay time is an abnormal stay time based on the first sub-recognition model;

In response to the current stay time being the abnormal stay time length, determining whether the characteristic information related to the vehicle stay meets the stay rule based on the second sub-recognition model;

In response to the characteristic information related to the vehicle stay that does not satisfy the stay rule, determining the abnormal evaluation value related to the current stay time based on the third sub-model;

Determine whether the abnormality evaluation value is greater than the abnormal stay threshold;

In response to the abnormality evaluation value being greater than the preset threshold value, it is determined that the vehicle is in an abnormal stay state.
The method according to claim 11, wherein the first sub-recognition model comprises a distribution-based Dukai's post-comparison model, and the first sub-recognition model based on determining whether the current stay duration is an abnormal stay duration includes :

Determine the identification coefficient based on the distribution-based Dukai's post-comparison model;

Determining at least one maximum anomaly estimated value based on the identification coefficient;

Determining whether the current stay time is greater than at least one of the at least one maximum abnormality estimation value;

In response to the current stay time being greater than at least one of the at least one maximum abnormality estimation value, it is determined that the current stay time length is the abnormal stay time length.
The method according to claim 11, wherein the characteristic information related to the stay of the vehicle includes at least one of the following: behavior of a person related to the vehicle, a current stay position of the vehicle; Determining whether the characteristic information related to the vehicle stay meets the stay rules, including:

Determining the stay rule based on the second sub-recognition model;

Determining whether the characteristic information related to the vehicle stay meets the stay rule includes:

Determine whether the behavior of the person related to the vehicle is a normal behavior; and/or

Determine whether the current stay position is in a hot spot area;

In response to that the behavior of the person related to the vehicle is not a normal behavior, and the current stay position is not in a hot spot area, it is determined that the characteristic information related to the stay of the vehicle does not satisfy the stay rule.
The method according to claim 13, wherein the third sub-recognition model comprises a density-based local abnormality factor algorithm LOF model; the third sub-model to determine the abnormality evaluation value related to the current stay time includes:

The local abnormality factor related to the current stay time is determined based on the local abnormality factor algorithm model, and the local abnormality factor is designated as the abnormality evaluation value.
The method of claim 11, wherein the at least one processor is further configured to:

In response to that the current stay time is not an abnormal stay time, it is determined that the vehicle is not in an abnormal stay state; or

In response to the feature information related to the stay of the vehicle satisfying the stay rule, it is determined that the vehicle is not in an abnormal stay state; or

In response to the abnormality evaluation value being less than or equal to the preset threshold value, it is determined that the vehicle is not in an abnormal stay state.
The method according to claim 10, wherein the real-time stay data includes current stay time and characteristic information related to vehicle stay, and based on the real-time stay data, a trained recognition model is used to determine whether the vehicle is abnormal Stay status, including:

Determining whether the current stay time is an abnormal stay time based on the first sub-recognition model;

Determining whether the characteristic information related to the vehicle stay meets the stay rule based on the second sub-recognition model;

Determining the abnormal evaluation value related to the current stay time based on the third sub-model and determining whether the abnormal evaluation value is greater than a preset threshold;

In response to the current stay time being the abnormal stay time, the characteristic information related to the vehicle stay does not meet the stay rules, and/or the abnormality evaluation value is greater than the preset threshold, it is determined that the vehicle is in an abnormal stay state .
The method of claim 10, wherein the method further comprises:

Acquire multiple training samples, the multiple training samples including first historical stay data of at least one first sample vehicle in the same vehicle state, and at least one second sample vehicle in the same time period and in the same area Second historical stay data; the first historical stay data and the second historical stay include historical stay duration and characteristic information related to historical vehicle stays,

Training a first initial sub-recognition model based on the first historical stay data to obtain a first sub-recognition model and recognition parameters related to the first sub-recognition model;

Training a second initial sub-recognition model based on the first historical stay data and the second historical stay data to obtain a second sub-recognition model and stay rules related to the second sub-recognition model;

Training a third initial sub-recognition model based on the second historical stay data to obtain a third sub-recognition model and an abnormal stay threshold related to the third sub-recognition model.
The method of claim 10, wherein the method further comprises:

Based on the real-time stay data, the recognition model is updated.
A computer-readable storage medium that stores computer instructions. After the computer reads the computer instructions in the storage medium, the computer executes the following abnormal stopping method for identifying a vehicle:

Acquiring real-time stay data of the vehicle;

Based on the real-time stay data, a recognition model is used to determine whether the vehicle is in an abnormal stay state; wherein the recognition model includes a machine learning model trained based on historical stay data of multiple historical vehicles.
A method for identifying abnormal staying behaviors of online car-hailing, which is characterized in that it includes:

Collect real-time data when a car-hailing stay occurs, the real-time data includes current stay location, current stay time, driver service status, and predetermined feature information;

Acquire a first historical data set and a second historical data set from a database; the first historical data set includes the historical stay time of at least one driver in the same service state as the real-time data, and the second historical data set includes The historical stay time of at least one driver in the same time period and in the same area as the real-time data;

According to the real-time data, the first historical data set, the second historical data set, and a preset model, identify whether the online car-hailing currently exhibits an abnormal staying behavior.
The method according to claim 20, wherein said identifying whether the online car-hailing is currently present based on the real-time data, the first historical data set, the second historical data set, and a preset model Abnormal stay behavior, including:

Input the current stay time length and the first historical data set into a first model, and determine whether the current stay time length is an abnormal stay time period relative to the first historical data set;

If it is determined that the current stay duration is the abnormal stay duration, input the predetermined characteristic information into the second model, and determine whether the predetermined characteristic information meets the preset rules of the second model;

If it is determined that the predetermined characteristic information does not meet the preset rules of the second model, the current stay time and the second historical data set are input into the third model to obtain the outliers of the current stay time The degree factor, if the outlier degree factor is greater than a preset threshold, it is determined that the online car-hailing currently exhibits an abnormal stay behavior.
The method according to claim 21, wherein the first model is a distribution-based Dukai's post-comparison model, and the coefficients used to represent the degree of abnormality in the Dukai's post-comparison model are preset coefficients;

The judging whether the current stay time is abnormal stay time relative to the first historical data set includes:

Obtaining the maximum estimated value of the first historical data set using the preset coefficient through the Du Kaifa post-comparison model;

If the current stay duration is greater than the maximum estimated value, it is determined that the current stay duration is the abnormal stay duration.
The method according to claim 22, wherein the first historical data set includes a first subset and a second subset; the first subset includes the current driver in the same service state as the real-time data Historical stay time; the second subset includes the historical stay time of all drivers in the database in the same service state and in the same area as the real-time data;

The judging whether the current stay time is abnormal stay time relative to the first historical data set includes:

Obtaining the first maximum estimated value of the first subset and the second maximum estimated value of the second subset using the preset coefficient through the Dukai's post-comparison model;

If the current stay duration is greater than the first maximum estimated value or the second maximum estimated value, it is determined that the current stay duration is an abnormal stay duration.
The method according to claim 23, wherein the predetermined characteristic information includes at least one of the following: passenger's payment behavior, passenger's evaluation of the driver, and billing hot spots;

The judging whether the predetermined characteristic information satisfies the preset rule of the second model includes:

Determine whether the payment behavior of the passenger is active payment; and/or

Determine whether the passenger’s evaluation of the driver is positive; and/or

Judging whether the current stay position is in the hot spot area for issuing bills;

If the foregoing judgment results are all no, it is determined that the predetermined characteristic information does not satisfy the preset rule of the second model.
The method according to any one of claims 21-24, wherein the third model is a local outlier factor (LOF) model based on density;

The obtaining the outlier degree factor of the current stay time includes:

According to the LOF model, the outlier degree factor LOF value of the current stay time is obtained according to the current stay time and the second historical data set.
The method according to claim 25, wherein the determining that the predetermined characteristic information does not satisfy the preset rule of the second model comprises:

If the number of samples in the second data set is less than the preset number, it is directly determined that the online car-hailing currently has an abnormal staying behavior;

The method also includes:

The number of historical stay data of all drivers in the database in the same time period as the real-time data is counted, and the result obtained by multiplying the number of historical stay data by a preset percentage is used as the preset number.
The method according to claim 21, further comprising:

If it is determined that the current stay duration is not the abnormal stay duration, it is determined that the online car-hailing does not currently exhibit any abnormal stay behavior; or

If it is determined that the predetermined characteristic information satisfies the preset rules of the second model, it is determined that there is no abnormal stay behavior in the online car-hailing; or

If it is determined that the outlier degree factor is less than or equal to the preset threshold, it is determined that the online car-hailing does not currently exhibit abnormal staying behavior.
A method for training an abnormal staying behavior recognition model for online car-hailing, which is characterized in that it includes:

Acquiring a training data set from a database, the training data including positive training data belonging to abnormal staying behaviors and negative training data not belonging to abnormal staying behaviors;

For each training data, extract the stay location, stay time, driver service status and predetermined feature information of the training data;

Acquire a first historical data set and a second historical data set from the database; the first historical data set includes the historical stay time of at least one driver in the same service state as the training data, and the second historical data The set includes the historical stay time of at least one driver in the same time period and in the same area as the training data;

According to the training data set, the first historical data set and the second historical data set corresponding to each training data, the preset model is trained to make the judgment result of the abnormal staying behavior output by the preset model accurate The rate and/or recall rate reach the target value.
The method according to claim 28, wherein the training of a preset model is performed according to the training data set, the first historical data set and the second historical data set corresponding to each training data ,include:

For each training data, input the stay duration and the first historical data set into a first model for training, and determine the preset coefficients of the first model;

Inputting the predetermined feature information into a second model for training, and determining a preset rule of the second model;

The staying time and the second historical data set are input into a third model for training, and a preset threshold value of the outlier degree factor is determined.
A device for identifying abnormal stay behavior of online car-hailing, which is characterized in that it comprises:

The collection module is used to collect real-time data when a car-hailing stay occurs, and the real-time data includes the current stay location, the current stay time, the driver's service status, and predetermined characteristic information;

The acquiring module is used to acquire a first historical data set and a second historical data set from a database; the first historical data set includes the historical stay time of at least one driver in the same service state as the real-time data, and the first historical data set 2. The historical data set includes the historical stay time of at least one driver in the same time period and in the same area as the real-time data;

The processing module is configured to identify whether the online car-hailing currently exhibits abnormal stay behavior according to the real-time data, the first historical data set, the second historical data set, and a preset model.
The device according to claim 30, wherein the processing module comprises:

The first processing module is configured to input the current stay duration and the first historical data set into a first model, and determine whether the current stay duration is abnormal stay duration relative to the first historical data set;

The second processing module is configured to input the predetermined characteristic information into the second model if it is determined that the current stay duration is the abnormal stay duration, and determine whether the predetermined characteristic information meets the preset rules of the second model;

The third processing module is configured to, if it is determined that the predetermined characteristic information does not meet the preset rules of the second model, input the current stay time and the second historical data set into the third model to obtain all The outlier degree factor of the current stay duration, and if the outlier degree factor is greater than a preset threshold, it is determined that the online car-hailing currently has an abnormal stay behavior.
The device according to claim 31, wherein the first model is a distribution-based Dukai's post-comparison model, and a coefficient used to represent the degree of abnormality in the Dukai's post-comparison model adopts a preset coefficient;

The first processing module is configured to: when determining whether the current stay duration is abnormal stay duration relative to the first historical data set:

Obtaining the maximum estimated value of the first historical data set by using the Dukai method post-comparison model with the preset coefficient;

If the current stay duration is greater than the maximum estimated value, it is determined that the current stay duration is the abnormal stay duration.
The device according to claim 32, wherein the first historical data set includes a first subset and a second subset; the first subset includes the current driver in the same service state as the real-time data Historical stay time; the second subset includes the historical stay time of all drivers in the database in the same service state and in the same area as the real-time data;

The first processing module is configured to: when determining whether the current stay duration is abnormal stay duration relative to the first historical data set:

Obtaining the first maximum estimated value of the first subset and the second maximum estimated value of the second subset using the preset coefficient through the Dukai's post-comparison model;

If the current stay duration is greater than the first maximum estimated value or the second maximum estimated value, it is determined that the current stay duration is an abnormal stay duration.
The device according to claim 31, wherein the predetermined characteristic information includes at least one of the following: passenger's payment behavior, passenger's evaluation of the driver, and billing hot spots;

The second processing module is configured to: when determining whether the predetermined feature information satisfies the preset rules of the second model:

Determine whether the payment behavior of the passenger is active payment; and/or

Determine whether the passenger’s evaluation of the driver is positive; and/or

Judging whether the current stay position is in the hot spot area for issuing bills;

If the foregoing judgment results are all no, it is determined that the predetermined characteristic information does not satisfy the preset rule of the second model.
The device according to any one of claims 31-34, wherein the third model is a density-based local abnormality factor (LOF) model;

When the third processing module obtains the outlier degree factor of the current stay time, it is used to:

According to the LOF model, the outlier degree factor LOF value of the current stay time is obtained according to the current stay time and the second historical data set.
The device according to claim 35, wherein after determining that the predetermined characteristic information does not satisfy the preset rule of the second model, the third processing module is further configured to:

If the number of samples in the second data set is less than the preset number, it is directly determined that the online car-hailing currently has an abnormal staying behavior;

The third processing module is also used for:

The number of historical stay data of all drivers in the database in the same time period as the real-time data is counted, and the result obtained by multiplying the number of historical stay data by a preset percentage is used as the preset number.
The device of claim 31, wherein:

The first processing module is further configured to, if it is determined that the current stay duration is not an abnormal stay duration, determine that there is no abnormal stay behavior in the online car-hailing currently; or

The second processing module is further configured to, if it is determined that the predetermined characteristic information satisfies the preset rules of the second model, determine that the online car-hailing does not currently exhibit abnormal stay behavior; or

The third processing module is further configured to, if it is determined that the outlier degree factor is less than or equal to a preset threshold, determine that there is no abnormal stay behavior in the online car-hailing currently.
An abnormal stay behavior recognition model training device for online car-hailing, which is characterized in that it comprises:

An obtaining module, configured to obtain a training data set from a database, the training data including positive training data belonging to abnormal staying behaviors and negative training data not belonging to abnormal staying behaviors;

The feature extraction module is used to extract the stay location, stay time, driver service status and predetermined feature information of the training data for each training data;

The acquiring module is further configured to acquire a first historical data set and a second historical data set from the database; the first historical data set includes the historical stay of at least one driver in the same service state as the training data Duration, the second historical data set includes the historical stay duration of at least one driver in the same time period and in the same area as the training data;

The training module is used to train the preset model according to the training data set, the first historical data set and the second historical data set corresponding to each training data, so that the abnormal output of the preset model stays The accuracy and/or recall rate of the behavior judgment result reaches the target value.
The device according to claim 38, wherein the training module is used to:

Input the length of stay and the first historical data set into a first model for training, and determine the preset coefficients of the first model; input the predetermined feature information into a second model for training, and determine the The preset rules of the second model; input the stay duration and the second historical data set into the third model for training, determine the preset threshold of the outlier degree factor, so that the output abnormal stay behavior judgment result The accuracy and/or recall rate reaches the target value.
A device for identifying abnormal stay behavior of online car-hailing, which is characterized in that it includes:

Memory

Processor; and

Computer program;

Wherein, the computer program is stored in the memory and is configured to be executed by the processor to implement the method according to any one of claims 20-27.
An equipment for training an abnormal staying behavior recognition model for online car-hailing, which is characterized in that it includes:

Memory

Processor; and

Computer program;

Wherein, the computer program is stored in the memory and is configured to be executed by the processor to implement the method according to claim 28 or 29.
A computer-readable storage medium, characterized in that a computer program is stored thereon;

When the computer program is executed by a processor, the method according to any one of claims 20-27 is implemented.
A computer-readable storage medium, characterized in that a computer program is stored thereon;

When the computer program is executed by a processor, the method according to claim 28 or 29 is realized.