CN110942220A

CN110942220A - Transport capacity scheduling method and device and server

Info

Publication number: CN110942220A
Application number: CN201811119726.1A
Authority: CN
Inventors: 常智华
Original assignee: Beijing Didi Infinity Technology and Development Co Ltd
Current assignee: Beijing Didi Infinity Technology and Development Co Ltd
Priority date: 2018-09-25
Filing date: 2018-09-25
Publication date: 2020-03-31
Anticipated expiration: 2038-09-25
Also published as: CN110942220B

Abstract

The application provides a capacity scheduling method, a capacity scheduling device and a server, and relates to the technical field of internet. Wherein, the method comprises the following steps: if an area with insufficient transport capacity in a specified time period is predicted, determining data containing silent driver information according to historical driver data corresponding to the area; wherein the silent drivers comprise off-line drivers with an active value within a set active value interval; screening data containing target driver information from the data containing the silent driver information; issuing a dispatch notice to the target driver to reach the zone within the specified time period. This application can follow and screen the target driver that can dispatch among the silence driver to dispatch target driver and go to the not enough region of power of transport, alleviate the passenger better and use the vigorous but not enough problem of driver supply with a car demand, effectively promoted user experience.

Description

Transport capacity scheduling method and device and server

Technical Field

The present application relates to the field of internet technologies, and in particular, to a capacity scheduling method, apparatus, and server.

Background

Along with the popularization of the taxi taking platform, more and more users choose to call the taxi through the taxi taking platform when going out, and a driver can receive orders through the taxi taking platform, so that services are provided for the users with taxi demands.

In the actual taxi taking process, due to the difference of user demands, the demand of the taxi is larger in some areas at a certain time, the demand of the taxi is smaller in another time, and a driver selects a taxi waiting place by experience, so that the phenomenon that the driver is not in supply and demand (namely, insufficient transport capacity) may occur in the areas in a peak time period, the taxi is difficult to call by the user, the taxi calling time is longer, and the user experience degree is poor.

Disclosure of Invention

In view of this, embodiments of the present application provide a capacity scheduling method, a capacity scheduling device, and a server, so as to solve the problems in the prior art that a passenger is difficult to call due to a driver's short supply and demand, and experience is poor.

In order to achieve the above purpose, the embodiment of the present invention adopts the following technical solutions:

in a first aspect, an embodiment of the present application provides a capacity scheduling method, where the method is applied to a platform server, and the method includes: if an area with insufficient transport capacity in a specified time period is predicted, determining data containing silent driver information according to historical driver data corresponding to the area; wherein the silent drivers comprise off-line drivers with an active value within a set active value interval; screening data containing target driver information from the data containing the silent driver information; issuing a dispatch notice to the target driver to reach the zone within the specified time period.

With reference to the first aspect, an embodiment of the present application provides a first possible implementation manner of the first aspect, where the method further includes: and predicting whether an area with insufficient transport capacity exists in a specified time period according to historical order data.

With reference to the first possible implementation manner of the first aspect, an embodiment of the present application provides a second possible implementation manner of the first aspect, where the step of predicting whether an area with insufficient transportation capacity exists in a specified time period according to historical order data includes: searching a hot spot area corresponding to a specified time period; calling at least one group of historical order data of the hot spot region corresponding to the specified time period;

counting the average value of unanswered orders in each group of the called historical order data; and if the average value exceeds a set first threshold value, determining the hot spot area as an area with insufficient transport capacity corresponding to the specified time period.

With reference to the second possible implementation manner of the first aspect, an embodiment of the present application provides a third possible implementation manner of the first aspect, where the step of retrieving at least one set of historical order data of the hotspot area corresponding to the specified time period includes: taking the current time as a reference, calling N sets of historical order data of the appointed time periods corresponding to the hot spot region to obtain N sets of historical order data; wherein N is a preset natural number.

With reference to the first aspect, an embodiment of the present application provides a fourth possible implementation manner of the first aspect, where the step of determining data including silent driver information according to historical driver data corresponding to the area includes: searching data containing candidate driver information of the area according to historical driver data; the historical driver data includes one or more of the following information: a history active area, a registered home address, a vehicle receiving place and a vehicle leaving place; and determining data containing silent driver information according to the online record of the data containing the candidate driver information in the specified time period.

With reference to the fourth possible implementation manner of the first aspect, an example of the present application provides a fifth possible implementation manner of the first aspect, where the step of searching for data including information of candidate drivers in the area according to historical driver data includes: performing GeoHash coding on the region according to longitude and latitude coordinates; acquiring an inverted index value corresponding to the region according to the GeoHash code corresponding to the region; determining data containing driver information corresponding to the inverted index value in an inverted index table as data containing candidate driver information corresponding to the region; the reverse index table is a corresponding relation table of reverse index values and driver information which is established in advance according to historical driver data.

With reference to the fourth possible implementation manner of the first aspect, this application example provides a sixth possible implementation manner of the first aspect, where the step of determining data containing silent driver information according to an online record of the data containing the candidate driver information in the specified time period includes: finding an online record of data containing the candidate driver information over the specified time period, the online record comprising: receiving and/or leaving records; and filtering the data containing the online driver information and the data containing the offline driver information, the activity value of which is lower than the lower limit value of the set activity value interval, in the data containing the candidate driver information to obtain the data containing the silent driver information.

With reference to the first aspect, this application provides a seventh possible implementation manner of the first aspect, where the step of filtering the data including the target driver information from the data including the silent driver information includes: screening data containing target driver information from the data containing silent driver information according to the insufficient degree of the transport capacity of the area; or screening data containing target driver information from the data containing the silent driver information according to the awakening probability of the silent driver.

With reference to the seventh possible implementation manner of the first aspect, this application provides an eighth possible implementation manner of the first aspect, where the step of screening the data containing the target driver information from the data containing the silent driver information according to the degree of the insufficient capacity of the area includes: determining the transport capacity gap number of the area according to the average value of unanswered orders of the area in the specified time period; counting the number of silent drivers; determining the number of target drivers according to the number of silent drivers and the number of the capacity gaps; and selecting the data containing the number of driver information from the data containing the silent driver information as the data containing the target driver information.

With reference to the seventh possible implementation manner of the first aspect, this application provides a ninth possible implementation manner of the first aspect, where the step of screening, according to the awakening probability of the silent driver, data containing target driver information from data containing silent driver information includes: calculating a wake-up probability of the silent driver based on a pre-trained machine learning model; and determining the data containing the silent driver information, of which the awakening probability is greater than a set second threshold value, as the data containing the target driver information.

With reference to the ninth possible implementation manner of the first aspect, an embodiment of the present application provides a tenth possible implementation manner of the first aspect, where the training process of the machine learning model includes: training a preset Xgboost model according to a historical awakening database of the silent driver; acquiring feature data output by the trained Xgboost model, and inputting the feature data into a preset LR model for training; and taking the trained LR model as the machine learning model.

With reference to the first aspect, an embodiment of the present application provides an eleventh possible implementation manner of the first aspect, where the step of issuing, to a target driver, a scheduling notification that the target driver reaches the area within the specified time period includes: acquiring a short message contact way of the target driver; and issuing a scheduling notice reaching the area within the specified time period to the target driver by using the short message contact way, wherein the scheduling notice comprises the identification of the area and the identification of the specified time period.

With reference to the first aspect, an embodiment of the present application provides a twelfth possible implementation manner of the first aspect, where the step of issuing, to a target driver, a scheduling notification that the target driver arrives at the area within the specified time period includes: acquiring a reward mechanism and scheduling information of the area; wherein the reward mechanism comprises: coupons and/or order draws that answer orders for the area are proportioned; the scheduling information includes: an identification of the region and an identification of the specified time period; issuing a dispatch notification to the target driver to reach the zone within the specified time period, the dispatch notification including the reward mechanism and the dispatch information.

In a second aspect, an embodiment of the present application further provides a capacity scheduling apparatus, where the apparatus is applied to a platform server, and the apparatus includes: the driver determining module is used for determining data containing silent driver information according to historical driver data corresponding to an area if an area with insufficient transport capacity in a specified time period is predicted; wherein the silent drivers comprise off-line drivers with an active value within a set active value interval; the driver screening module is used for screening data containing target driver information from the data containing the silent driver information; and the driver scheduling module is used for issuing a scheduling notice reaching the area within the specified time period to the target driver.

In combination with the second aspect, embodiments of the present application provide a first possible implementation manner of the second aspect, where the apparatus further includes: and the area prediction module is used for predicting whether an area with insufficient transport capacity exists in a specified time period according to historical order data.

With reference to the first possible implementation manner of the second aspect, this application provides a second possible implementation manner of the second aspect, where the region prediction module is configured to: searching a hot spot area corresponding to a specified time period; calling at least one group of historical order data of the hot spot region corresponding to the specified time period; counting the average value of unanswered orders in each group of the called historical order data; and if the average value exceeds a set first threshold value, determining the hot spot area as an area with insufficient transport capacity corresponding to the specified time period.

With reference to the second possible implementation manner of the second aspect, this application provides a third possible implementation manner of the second aspect, where the region prediction module is configured to: taking the current time as a reference, calling N sets of historical order data of the appointed time periods corresponding to the hot spot region to obtain N sets of historical order data; wherein N is a preset natural number.

In combination with the second aspect, the present application provides a fourth possible implementation manner of the second aspect, wherein the driver determination module is configured to: searching data containing candidate driver information of the area according to historical driver data; the historical driver data includes one or more of the following information: a history active area, a registered home address, a vehicle receiving place and a vehicle leaving place; and determining data containing silent driver information according to the online record of the data containing the candidate driver information in the specified time period.

In combination with the fourth possible implementation manner of the second aspect, the present application provides a fifth possible implementation manner of the second aspect, wherein the driver determination module is configured to: performing GeoHash coding on the region according to longitude and latitude coordinates; acquiring an inverted index value corresponding to the region according to the GeoHash code corresponding to the region; determining data containing driver information corresponding to the inverted index value in an inverted index table as data containing candidate driver information corresponding to the region; the reverse index table is a corresponding relation table of reverse index values and driver information which is established in advance according to historical driver data.

In combination with the fourth possible implementation manner of the second aspect, the present application provides a sixth possible implementation manner of the second aspect, wherein the driver determination module is configured to: finding an online record of data containing the candidate driver information over the specified time period, the online record comprising: receiving and/or leaving records; and filtering the data containing the online driver information and the data containing the offline driver information, the activity value of which is lower than the lower limit value of the set activity value interval, in the data containing the candidate driver information to obtain the data containing the silent driver information.

With reference to the second aspect, the present application provides a seventh possible implementation manner of the second aspect, wherein the driver screening module is configured to: screening data containing target driver information from the data containing silent driver information according to the insufficient degree of the transport capacity of the area; or screening data containing target driver information from the data containing the silent driver information according to the awakening probability of the silent driver.

With reference to the seventh possible implementation manner of the second aspect, the present application provides an eighth possible implementation manner of the second aspect, wherein the driver screening module is configured to: determining the transport capacity gap number of the area according to the average value of unanswered orders of the area in the specified time period; counting the number of silent drivers; determining the number of target drivers according to the number of silent drivers and the number of the capacity gaps; and selecting the data containing the number of driver information from the data containing the silent driver information as the data containing the target driver information.

With reference to the seventh possible implementation manner of the second aspect, the present application provides a ninth possible implementation manner of the second aspect, wherein the driver screening module is configured to: calculating a wake-up probability of the silent driver based on a pre-trained machine learning model; and determining the data containing the silent driver information, of which the awakening probability is greater than a set second threshold value, as the data containing the target driver information.

With reference to the ninth possible implementation manner of the second aspect, the present application provides a tenth possible implementation manner of the second aspect, where the training process of the machine learning model includes: training a preset Xgboost model according to a historical awakening database of the silent driver; acquiring feature data output by the trained Xgboost model, and inputting the feature data into a preset LR model for training; and taking the trained LR model as the machine learning model.

In combination with the second aspect, the present application provides an eleventh possible implementation manner of the second aspect, wherein the driver dispatching module is configured to: acquiring a short message contact way of the target driver; and issuing a scheduling notice reaching the area within the specified time period to the target driver by using the short message contact way, wherein the scheduling notice comprises the identification of the area and the identification of the specified time period.

In combination with the second aspect, this embodiment provides a twelfth possible implementation manner of the second aspect, where the driver dispatching module is configured to: acquiring a reward mechanism and scheduling information of the area; wherein the reward mechanism comprises: coupons and/or order draws that answer orders for the area are proportioned; the scheduling information includes: an identification of the region and an identification of the specified time period; issuing a dispatch notification to the target driver to reach the zone within the specified time period, the dispatch notification including the reward mechanism and the dispatch information.

In a third aspect, a server comprises a memory and a processor, wherein the memory is used for storing a program supporting the processor to execute the method in the first aspect or any one of the first to twelfth possible implementation manners of the first aspect.

In a fourth aspect, the present application further provides a computer-readable storage medium for storing computer software instructions for an apparatus in the second aspect or any one of the first to twelfth possible implementation manners of the second aspect.

According to the capacity scheduling method, the capacity scheduling device and the capacity scheduling server, data containing silent driver information can be determined according to historical driver data corresponding to a capacity shortage area, the data containing target driver information are screened from the data containing the silent driver information, and then a scheduling notice that the capacity shortage area is reached in a specified time period is sent to a target driver. This kind of select the target driver that can dispatch from silent driver to dispatch target driver and go to the regional mode that the capacity is not enough, can alleviate passenger better and use the vigorous but insufficient supply of driver's demand problem of car, effectively promoted user experience.

In order to make the aforementioned objects, features and advantages of the embodiments of the present application more comprehensible, embodiments accompanied with figures are described in detail below.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are required to be used in the embodiments will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present application and therefore should not be considered as limiting the scope, and for those skilled in the art, other related drawings can be obtained from the drawings without inventive effort.

Fig. 1 is a schematic view illustrating an application scenario of a taxi taking platform provided in an embodiment of the present application;

fig. 2 is a flowchart illustrating a capacity scheduling method provided in an embodiment of the present application;

FIG. 3a is a schematic diagram illustrating a screening of a target driver provided by an embodiment of the present application;

FIG. 3b is a schematic diagram illustrating another example of a target driver screening provided by an embodiment of the present application;

FIG. 4 is a flowchart illustrating a specific method for finding candidate drivers according to an embodiment of the present application;

FIG. 5 is a schematic diagram illustrating region coding provided in an embodiment of the present application;

FIG. 6 is a flowchart illustrating a method for training a machine learning model according to an embodiment of the present disclosure;

fig. 7 is a flowchart illustrating another capacity scheduling method provided in an embodiment of the present application;

fig. 8 is a block diagram illustrating a structure of a capacity scheduling apparatus according to an embodiment of the present application;

fig. 9 is a block diagram illustrating a structure of another capacity scheduling apparatus provided in an embodiment of the present application;

fig. 10 shows a schematic structural diagram of a server provided in an embodiment of the present application.

Detailed Description

In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all the embodiments. The following detailed description of the embodiments of the present application is not intended to limit the scope of the claimed application, but is merely representative of selected embodiments of the application. All other embodiments, which can be derived by a person skilled in the art from the embodiments of the present application without making any creative effort, shall fall within the protection scope of the present application.

The method, the device, the electronic equipment or the computer storage medium in the embodiment of the application can be applied to any scene where the taxi taking platform needs to carry out capacity dispatching (namely, dispatching a driver). The embodiments of the present application do not limit specific application scenarios, and any scheme for performing capacity scheduling by using the method provided by the embodiments of the present application is within the scope of the present application.

First, referring to an application scenario diagram of a taxi taking platform shown in fig. 1, a platform server (specifically, a server of the taxi taking platform) and a passenger terminal and a driver terminal respectively connected to the platform server in a communication manner are illustrated in fig. 1. The passenger terminal may be a mobile terminal such as a mobile phone of a passenger, and the driver terminal may be a mobile terminal such as a mobile phone of a driver and an ipad, and may also be a vehicle-mounted device installed in a driver vehicle.

The passenger terminal and the driver terminal are respectively provided with a passenger client (or called as a passenger end APP) and a driver client (or called as a driver end APP) of the taxi taking platform. The passenger who has the demand for calling can input information such as a starting place and a destination at the passenger side APP, the passenger side APP generates an order and sends the order to the platform server, the platform server can send the passenger order to the driver side APP which meets the order receiving condition (such as being close to the starting place of the passenger), and the driver who receives the order and confirms the order receiving provides service for the passenger.

However, there is a class of drivers on the taxi taking platform, and although the drivers are registered drivers, the drivers may not log in the taxi taking platform for a long time (i.e. in an off-line state) or the drivers log in the taxi taking platform but the order taking frequency is low, in the embodiment of the present application, the drivers who are not currently on-line, are more active drivers (the drivers may not be currently on-line due to other situations), and the like, have the potential of taking orders, but the drivers who are not currently in the order taking state are called silent drivers. That is, the silent drivers mentioned in the embodiments of the present application include off-line drivers with active values in the active value setting interval. An off-line driver is a driver who is not currently logged in the taxi taking platform. Such drivers usually have a certain activity level on the taxi taking platform, but the activity level is lower than that of the drivers who normally take the taxi, so that the current activity value of each driver can be counted according to historical driver data, and if the activity value is within a set activity value interval and the driver is not on line currently, the driver can be regarded as a silent driver. The statistical mode of the active value can be to select one or more parameters from the parameters of the online time length, the registered time length, the number of the order taking, the running mileage of the order taking and the like of the driver for statistics, the statistical historical driver data can also be the data in the specified time length before the current time, and the specified time length can be set according to experience, such as one week or two weeks, or one month or two months and the like. The specific statistical process of the active value can be flexibly formulated according to needs, and the embodiment of the invention does not limit the process.

The capacity scheduling method, the capacity scheduling device and the capacity scheduling server are provided for solving the problems that passengers are difficult to call vehicles in hot spot areas or peak periods and drivers are not in supply and demand. This is illustrated in detail by the following examples.

Example 1

Referring to fig. 2, a flowchart of a capacity scheduling method, which may be applied to a platform server, includes the following steps:

step S202, if an area with insufficient transport capacity in a specified time period is predicted, determining data containing silent driver information according to historical driver data corresponding to the area. Here, the data containing the silent driver information may be simply referred to as silent driver data.

In practical applications, the specified time period may also be a period of time in the future from the present, such as 1 hour in the future or 30 minutes in the future from the present; of course, the specified time period may also be a peak time period of the vehicle utilization counted by the server, such as daily on-duty and off-duty time periods of 7: 00-8: 00, 11: 00-13: 00 and 17: 00-20: 00, or a full day of a holiday. In one embodiment, the designated time period may be automatically determined by the server based on historical booking needs, and in another embodiment, the designated time period may be manually set.

The areas with insufficient transportation capacity are areas which are not in demand of drivers, such as office building gathering areas, school doorways, residential areas, shopping mall gathering areas and the like. In these areas of insufficient capacity, the number of drivers on the taxi platform that can receive orders is usually less than the number of passenger orders, and it is difficult to meet the passenger demand for taking the car.

It is understood that the server will generally record the historical data of each driver, such as the historical active area of the driver, the registered home address, the daily vehicle pickup location, the daily vehicle pickup time, the daily active time length, and the like, and specifically refer to one form of the historical driver data shown in table 1, and the historical driver data of the embodiment of the present application is not limited to the form of a table, and may be in other data formats.

TABLE 1

The table 1 simply shows the historical data of two drivers, wherein the daily vehicle receiving place and the daily vehicle leaving place, the daily vehicle receiving time and the daily vehicle leaving time can be set by the driver through a driver side APP, or can be obtained by a server through statistical analysis of the historical driving track of Zhang III; the daily activity duration may be the driver's daily on-line duration or the operating duration. For example, Zhang III is a designated driver, the daily departure place of Zhang III is Zhang III residential area (X area) and near office building (Y area), and the daily departure time is the departure time of Zhang III (8: 00 am, 20:00 pm). The LiIV is a full-time taxi driver, the daily taxi taking-out places of the LiIV are unified into a taxi company (Z zone), the daily taxi taking-out time is 14:00 in the afternoon, the daily taxi taking-out time is 23:00 in the evening and the like.

The server can determine data (which can be referred to as candidate driver data for short) containing the information of the candidate driver corresponding to the area with insufficient transport capacity according to the area with insufficient transport capacity and the statistical historical data of each driver, and further screen the data containing the information of the silent driver from the candidate driver data. For example, the area with insufficient transport capacity is an area A, the vehicle receiving/sending place of Zhang III comprises the area A, the historical active area of Li IV is the area A, and then Zhang III and Li IV are candidate drivers of the area A. But the online time length of Zhang three and Li four and historical order taking data show that: three lie's basically connect the single on line every day, and the active value is very high, and lie four is short online duration, and it is also less to connect single quantity, then the server considers lie four to be the silence driver that A district corresponds.

In step S204, data including the target driver information is selected from the data including the silent driver information. The data containing the target driver information can be referred to as target driver data for short, and the target driver is also a dispatchable driver.

The method for screening target driver data from silent driver data includes selecting a certain amount of driver data randomly according to the degree of insufficient transportation capacity when the number of the found silent drivers is large, and using the found silent driver data as the target driver data when the number of the found silent drivers is small. In addition, the server can calculate the awakening probability of each silent driver by adopting a mode such as machine learning, and the silent driver data with the awakening probability higher than a preset threshold value is determined as the target driver data.

For ease of understanding, referring to a screening schematic of target drivers as shown in fig. 3a, it is shown that the server determines a plurality of silent drivers associated with the area of insufficient capacity from historical driver data, and then collects a certain number of target drivers from the silent drivers, and the collection range may depend on the degree of insufficient capacity. The circle taking mode is also various, and can be concentrated circle taking or dispersed circle taking. Referring to fig. 3b, another screening diagram of target drivers, illustrating the wake-up probability of each silent driver, fig. 3b identifies silent drivers with wake-up probability higher than 60% as target drivers.

Step S206, issuing a dispatching notice reaching the area within a specified time period to the target driver. The above-mentioned area is also an area with insufficient transport capacity. In specific implementation, the contact information of the target driver can be acquired, specifically, the contact information is preferably a short message contact information, and with the development of technology, a communication method similar to a third party contact information such as WeChat and QQ can be adopted, and certainly, a voice contact information can also be included. And then using the contact means to send a scheduling notice to the target driver, wherein the scheduling notice reaches the area with insufficient transport capacity within the specified time period, and comprises the identification of the area with insufficient transport capacity and the identification of the specified time period. For example, a short message "Mr. Zhang Sanqi is good" is sent to a target driver, the vehicle demand in the area A is expected to be large in the period of 18: 00-20: 00, and the driver is advised to go to the area A to pick up a bill at that time ".

In order to prompt the target driver to actively respond to the platform scheduling, the server can acquire the reward mechanism and scheduling information of the area with insufficient capacity; wherein, the reward mechanism can comprise a coupon and/or an order drawing proportion for answering the order of the area, etc., the scheduling information can comprise the identification of the area, the identification of the designated time period, etc., and then the target driver is issued a scheduling notice for reaching the area with insufficient transport capacity in the designated time period, and the scheduling notice comprises the reward mechanism and the scheduling information. For example, a target driver is sent a message of three times, namely that Mr. three times you, such as 18: 00-20: 00, meets the bill in the area A, and each bill can enjoy 10 yuan of bonus. By the method, silent drivers can be effectively awakened, the silent drivers are encouraged to cooperate with dispatching, and the problem that the drivers in the hot spot area are short of supply and demand is solved.

According to the capacity scheduling method provided by the embodiment of the application, the data containing the silent driver information can be determined according to the historical driver data corresponding to the capacity shortage area, the data containing the target driver information is screened from the data containing the silent driver information, and then the scheduling notice reaching the capacity shortage area in the specified time period is sent to the target driver. This kind of select the target driver that can dispatch from silent driver to dispatch target driver and go to the regional mode that the capacity is not enough, can alleviate passenger better and use the vigorous but insufficient supply of driver's demand problem of car, effectively promoted user experience.

When the transportation capacity scheduling method is specifically implemented, it is necessary to first determine whether an area with insufficient transportation capacity exists in an area covered by a taxi taking platform. In one embodiment, whether an area of insufficient capacity exists within a specified time period may be predicted from historical order data. The platform server may perform statistical analysis on the historical orders, such as determining order generation quantity and order response quantity corresponding to each area, and if the order response quantity of a certain area is lower than the order generation quantity, it indicates that the area is an area with insufficient transportation capacity, and the method may specifically refer to the following steps:

(1) and searching a hot spot region corresponding to the specified time period.

It can be understood that the hotspot regions corresponding to different time periods are different. For example: if the time is in the working/school hours (such as 7: 00-9: 00), the hot spot area can be a residential area and the like; if the time is in the period of leaving work/leaving school (such as 17: 00-20: 00), the hot spot area can be an area concentrated in office buildings or a school and the like; if the mobile terminal is not in working hours, the hotspot area can also be an area where a shopping mall is located, an area where tourist attractions are located, and the like; in addition, public areas such as train stations, airports, business districts, etc. may be hot spot areas regardless of time periods. In a specific implementation, the hotspot region may be obtained by the platform server in the following manner: counting the order quantity of the target area in a preset designated time period; and setting the target area with the order quantity exceeding the set threshold as the hot spot area in the specified time period. Or, the hot spot area corresponding to the specified time period may also be preset manually, for example, an area correspondence table is manually established according to data of the platform server or third-party data, a correspondence between each time period and the hot spot area is stored in the area correspondence table, and the platform server may search the hot spot area corresponding to the specified time period according to the area correspondence table.

(2) And calling at least one group of historical order data of the hot spot region corresponding to the specified time period.

In one embodiment, the current time can be taken as a reference, and N sets of historical order data corresponding to the hot spot region in N designated time periods are called to obtain N sets of historical order data; wherein N is a preset natural number. For example, if the hot spot area is an area A concentrated in an office building and the designated time period corresponding to the area is 18: 00-21: 00, historical order data of 18: 00-21: 00 every day in the previous 30 days can be called by taking the current time as a reference, and 30 groups of historical order data are obtained. Each set of historical order data may include an order generated quantity, an order answered quantity, and/or an order unanswered quantity within a specified time period.

(3) And counting the average value of the unanswered orders in each set of the called historical order data. By calculating the average value of the unanswered orders of each group of historical order data, special conditions can be effectively filtered, and whether the phenomenon that a driver is short of supply and demand occurs in a hot spot area in a specified time period or not can be objectively reflected.

(4) And if the average value exceeds a set first threshold value, determining the hot spot area as an area with insufficient transport capacity corresponding to the designated period. For example, if the average value of the unanswered orders in the area A in the designated time period is 50 orders and exceeds the set 30 orders, the area A is an area with insufficient transport capacity corresponding to the time period of 18: 00-21: 00, and if the average value of the unanswered orders in the area A in the designated time period is 25 orders and is lower than the set 30 orders, the area A does not belong to the area with insufficient transport capacity corresponding to the time period of 18: 00-21: 00.

By the mode, the area with insufficient capacity corresponding to the designated time period can be objectively and effectively determined, and the accuracy and the reliability of capacity scheduling are further improved.

Since a large number of drivers are registered with the taxi-taking platform, in order to determine silent drivers therefrom, in one embodiment, data containing candidate driver information for a region can be searched based on historical driver data; and then determining the data containing the silent driver information according to the online record of the data containing the candidate driver information in a specified time period. Wherein the historical driver data may include one or more of the following: historical active areas, registered home addresses, pickup locations, departure locations, etc. Referring to fig. 4 in detail, a flowchart of a specific method for finding a candidate driver includes the following steps:

and step S402, performing GeoHash coding on the area (namely the area with insufficient transport capacity) according to the longitude and latitude coordinates. For ease of understanding, the GeoHash code will first be explained in detail:

the GeoHash algorithm is an address coding method, which can code two-dimensional space longitude and latitude data into a one-dimensional character string, such as that the GeoHash code of the north sea park is wx4g0ec 1. The basic principle is to understand the earth as a two-dimensional plane, and recursively decompose the plane into smaller sub-blocks (i.e., grid the plane), each sub-block having the same code in a certain latitude and longitude range. Referring to a schematic diagram of region coding shown in fig. 5, GeoHash codes corresponding to 9 regions of a certain city are represented as WX4ER, WX4G2, WX4G3, WX4EP, WX4G0, WX4G1, WX4DZ, WX4FB, and WX4 FC. It is understood that the GeoHash codes of different regions differ. In practical application, the regions can be divided according to requirements, and the longer the GeoHash code is, the more accurate the represented region range is. For example, a 5-bit code can represent a rectangular area in the range of 10 square kilometers, while a 6-bit code can represent a finer area (about 0.34 square kilometers). Similar codes indicate that the regions are close in distance, such as regions corresponding to WX4G2 and regions corresponding to WX4G3 being similar regions.

According to the longitude and latitude coordinates corresponding to the area with insufficient transport capacity, the GeoHash code corresponding to the area can be determined by searching a preset area code table. In particular, the Search may be performed by a third-party Search server, such as a Search engine ES (Elastic Search), to realize a real-time, fast and stable Search.

Step S404, obtaining an inverted index value corresponding to the region according to the GeoHash code corresponding to the region, and determining data containing driver information corresponding to the inverted index value in an inverted index table as data containing candidate driver information corresponding to the region; the reverse index table is a corresponding relation table of reverse index values and driver information which is established in advance according to historical driver data.

The inverted index table can firstly acquire the associated GeoHash codes of the drivers according to historical driver data. Specifically, the associated GeoHash code of the driver is the GeoHash code of the associated area of the driver. The driver's associated area is an area related to the driver, such as a daily boarding place, a daily departure place, a history active place of the driver, a registered home address of the driver, and the like, of the driver can be considered as the driver's associated area. Therefore, the driver association region can be obtained through the historical driver data counted by the platform server, and then the GeoHash code (namely, the associated GeoHash code) of the driver association region can be determined through the region code table, then the reverse index value corresponding to the code is obtained, and then the corresponding relation between the driver information and the reverse index value is stored in the reverse index table.

Through the candidate driver searching mode based on the GeoHash code, driver data related to an area with insufficient transport capacity can be effectively searched, after the candidate driver data of the area with insufficient transport capacity are determined, an online record of the candidate driver data in a specified time period can be searched, wherein the online record can comprise: receiving and/or leaving records; of course, the online time length, the driving record, the order taking record and the like can also be included. Then filtering out the online driver data in the candidate driver data to obtain the offline driver data, wherein all the offline drivers can be regarded as silent drivers, or the range of the silent drivers can be further narrowed to improve the effectiveness of dispatching, the offline driver data with the activity value lower than the lower limit value of the set activity value interval are filtered out from the offline driver data, and the rest driver data are used as silent driver data. For example, if the candidate drivers in the area-A with insufficient transport capacity are determined to be Zhang III and Li IV respectively, and Zhang III records the vehicle receiving time in the specified time period of 18: 00-21: 00 or has a certain online time length; and since the fourth plum basically has no online record in the specified time period of 18: 00-21: 00, the third plum can be determined to be an online driver, and the fourth plum is determined to be a silent driver.

After determining the silent driver data, the embodiment further provides the following three specific implementations of screening target driver data from the silent driver data:

the first method is as follows: and screening target driver data from the silent driver data according to the insufficient capacity degree of the area. That is, schedulable target driver data is screened from silent driver data according to the degree of driver scarcity.

In specific implementation, the following steps can be referred to:

(1) and determining the transport capacity gap number of the area according to the average value of the unanswered orders of the area in the specified time period. That is, by the server counting the average value of unanswered orders in a specified period in an area with insufficient capacity, the missing driver number in the area, which is usually in the specified period, can be predicted. In one embodiment, the capacity gap number is equal to the average unanswered order; in another embodiment, the number of capacity gaps is equal to the product of the unresponsive average and a predetermined factor (less than 1).

For example, if the average value of unanswered orders in the area A in the specified time period of 18: 00-21: 00 is 50 orders, the number of transportation gaps in the area is 50 drivers; or the conveying capacity gap number of the area is 40-50 x 0.8; wherein 0.8 is a preset coefficient. The reason for setting the preset coefficient is to take into account that a part of the drivers can complete a plurality of orders within a specified period. The driver may complete multiple orders one by one or simultaneously. For example, in a car-sharing mode provided by a car-parking platform, if a plurality of passengers with the car-sharing mode selected and the routes approximately the same place orders, the drivers can simultaneously pick up the car-sharing passengers to complete a plurality of orders at one time. In practical application, the coefficient can be determined according to the historical order quantity and the historical order receiving quantity, and then the coefficient is multiplied by the average value of the unanswered orders, so that the driver quantity required by completing the average value of the unanswered orders can be determined, and the required driver quantity is also the capacity gap number of the area.

(2) The number of silent drivers is counted. For example, candidate driver data related to an area with insufficient transport capacity is determined through a GeoHash code, silent driver data are screened according to online records of the candidate driver data such as vehicle receiving records, online time length and order receiving records in a specified time period, and then the number of silent drivers is counted.

(3) And determining the number of target drivers according to the number of silent drivers and the number of capacity gaps. In one embodiment, if the number of silent drivers is less than the number of capacity gaps, the number of target drivers is determined as the number of silent drivers, that is, all silent drivers are used as target drivers. In another embodiment, the number of target drivers is determined from the number of silent drivers to be equal to the number of capacity gaps if the number of silent drivers is greater than the number of capacity gaps.

(4) And selecting the driver data of the quantity from the silent driver data as target driver data. After the number of the target drivers is determined, the number of the drivers can be randomly selected from the silent drivers to be the schedulable target drivers, or the degree of silence of each silent driver can be evaluated, then the silent drivers are reordered according to the degree of silence, and then the number of the drivers from front to back is selected to be the target drivers.

The second method comprises the following steps: and screening target driver data from the silent driver data according to the awakening probability of the silent driver. Namely, by evaluating the awakening probability of each silent driver, the driver with higher awakening probability is screened from the silent drivers as the target driver. Wherein the wake-up probability is the probability that a silent driver is willing to accept platform scheduling.

In specific implementation, the following steps can be referred to:

(1) the wake-up probability of a silent driver is calculated based on a pre-trained machine learning model. The machine learning model is a key technology of artificial intelligence at present, and can analyze and process data and obtain a required result. Such as a logistic regression model, a random forest model, a bayesian method model, a support vector machine model, a neural network model, etc., belong to machine learning models. If the machine learning model is to be put into practical application, the machine learning model needs to be trained by adopting a training data set, so that the machine learning model meets the requirements through continuous training of supervised learning or unsupervised learning, and the required result is accurately output. In this embodiment, a neural network model may be used as a machine learning model for training, and the wake-up probability of a silent driver is calculated.

(2) And determining the silent driver data with the awakening probability larger than a set second threshold value as the target driver data. For example, silent driver data with a wake-up probability greater than 50% is determined as the target driver data.

The target driver is screened through the awakening probability, the possibility that the target driver receives dispatching can be improved, and the transport capacity dispatching effect of the taxi taking platform is guaranteed.

This embodiment further provides a training process of a machine learning model, referring to a flowchart of a training method of a machine learning model shown in fig. 6:

step S602, training a preset XGBoost model according to a historical awakening database of the silent driver.

The XGboost (eXtreme Gradient Boosting) model is a prediction model with a better current effect, and in practical application, training data can be obtained from a historical awakening database of a silent driver, and then the XGboost model is trained by adopting the training data. The historical awakening database can store relevant information of the silent driver and a corresponding awakening result; the relevant information of the silent driver can comprise the offline duration, the historical order taking amount and the like of the silent driver. The wake-up result may be whether the silent driver accepts the platform dispatch, or the wake-up result may be the number of times the silent driver accepts the dispatch in the historical dispatch process, etc. The historical wake-up database may take the form of data shown in table 2; wherein, the related information of the driver in the recent 1 month is counted in table 2.

TABLE 2

As can be seen from Table 2, 480h of Zhang three drivers are in an off-line state within one month, and only the order is received 10 times, but the history records of the coordinated dispatching are recorded, and 2 orders belong to the coordinated dispatching order in the 10 orders. The probability of cooperative scheduling may be a ratio of the number of times of cooperative scheduling of zhang san to the number of times of scheduling notification sent by the platform server to zhang san, for example, zhang san receives 20 scheduling notifications sent by the platform server in a month, and zhang san responds only 2 times in a month, so the probability of cooperative scheduling of zhang san in 1 month is 2/20 ═ 10%.

The XGboost model can be trained through the data in the historical wake-up database, so that the XGboost model outputs a result meeting the requirements of the embodiment.

Step S604, acquiring feature data output by the trained XGBoost model, and inputting the feature data into a preset LR model for training. In specific implementation, the leaf node obtained after the training data is input to the XGboost model may be used as the feature data output by the XGboost model, and then the LR (Logistic Regression) model is trained by using the leaf node (degree is 0, also called terminal node). The LR model is mainly characterized in that a logic function is applied on the basis of linear regression, the fitting degree and the model interpretation degree of data can be better considered, the calculation is simple and convenient, and the accurate awakening probability of the silent driver can be finally obtained. Specific structures of the XGboost model and the LR model can be implemented with reference to related technologies, and will not be described herein.

In step S606, the trained LR model is used as a machine learning model.

According to the embodiment of the application, the XGboost and LR fusion mode is adopted to obtain the machine learning model, the characteristics and advantages of the XGboost model and the LR model are utilized, the machine learning model required for calculating the awakening probability is obtained through training, and the platform server can accurately and reliably evaluate the awakening probability of each silent driver by means of the machine learning model.

The third method comprises the following steps: and screening target driver data according to the number of the searched silent driver data. That is, the screening process for the target driver is largely dependent on the number of silent drivers.

In specific implementation, the following steps can be referred to:

(1) and judging whether the number of the searched silent driver data is higher than a preset number. This way the number of silent drivers is measured by a preset number. The preset number can be set manually or set by the server according to historical experience.

(2) If yes, randomly selecting a certain amount of target driver data according to the insufficient capacity degree. That is, when the number of the found silent drivers is large, the schedulable target driver is randomly selected from the large number of silent drivers according to the number of the missing drivers.

(3) And if not, taking the found silent driver data as target driver data. That is, when the number of found silent drivers is small, all silent drivers are taken as schedulable target drivers.

In practical application, any one of the first to third modes can be flexibly selected according to requirements, or multiple modes of the three modes can be combined, such as comprehensively screening target drivers according to the insufficient degree of the transport capacity, the awakening probability of the drivers and the number of searched silent drivers. Of course, the above is only an exemplary illustration, and the screening method thereof may also be adopted, which is not described herein again.

By the aid of the capacity scheduling method, silent driver data can be determined according to historical driver data corresponding to the capacity shortage area, target driver data can be screened from the silent driver data, and then the target driver can be scheduled. This kind of mode can alleviate the passenger better and use the vigorous but insufficient problem of driver supply of car demand, when having reduced passenger's the time of taking a bus on the one hand, on the other hand also brings the income for the driver, has comprehensively promoted user experience.

Example two

In combination with the first embodiment, this embodiment provides a specific capacity scheduling method, which may be applied to a platform server, and the platform server may dispatch a schedulable driver among silent drivers to a required area to receive an order through the capacity scheduling method provided by this embodiment. Referring to another flow chart of the capacity scheduling method shown in fig. 7, the method specifically includes the following steps:

in step S702, an area with insufficient capacity existing in a specified time period is predicted from the historical order data.

In specific implementation, the platform server searches for a hot spot area within a specified time period (such as 1 hour in the future) by performing statistical analysis on historical order data, then determines an average value of unanswered orders of the hot spot area according to the historical order data of the hot spot area, and further determines the hot spot area with the average value of answered orders higher than a set first threshold value as an area with insufficient transport capacity. In practical applications, the historical order data may specifically include an order starting and ending point, an order starting and ending time and the like, the historical order data about the passenger may further include a daily getting-on and getting-off time of the passenger, a daily order track and a historical order quantity of the passenger, and the historical order data about the driver may further include a daily getting-on and getting-off place, a getting-on time, a getting-off time, a historical getting-on and getting-off quantity of the driver, a driving track rule of the driver and the like.

And step S704, determining candidate driver data of the region according to the GeoHash code.

In specific implementation, the GeoHash codes of the transport capacity deficiency areas can be obtained through longitude and latitude coordinates corresponding to the transport capacity deficiency areas, and the associated GeoHash codes of drivers are obtained according to historical driver data, wherein the associated GeoHash codes are the GeoHash codes of the associated areas of the drivers; if the driver's associated GeoHash code has a GeoHash code of an area of insufficient transport, the driver is determined to be a candidate driver for the area of insufficient transport. In this way, drivers associated with areas of insufficient capacity, all of which may be considered regional capacity, are effectively and accurately known.

Step S706, online driver data in the candidate driver data is filtered through the vehicle receiving record and/or the vehicle leaving record, and silent driver data and offline driver data with the activity value lower than the lower limit value of the set activity value interval are obtained.

In specific implementation, candidate drivers can be screened according to the vehicle receiving records of the drivers, online drivers and offline drivers with the activity values lower than the lower limit value of the set activity value interval are filtered, and silent drivers are determined. Specifically, according to the departure/reception status of the driver, the current order taking situation of the driver can be judged, so as to identify a silent driver, wherein the silent driver can be a driver who does not take the order any more or has taken the line off in a specified time period.

In step S708, the wake-up probability of each silent driver is calculated by the pre-trained machine learning model.

In specific implementation, the machine learning model may be an XGBoost + LR fusion model, and the structure thereof may be implemented by referring to related technologies, which is not described herein again. The awakening probability of the silent driver can be obtained by inputting the historical driver data of the silent driver into a machine learning model (also called an awakening model) trained in advance.

Step S710, determining target driver data which can be called according to the awakening probability of each silent driver.

Silent drivers with a wake-up probability greater than a set second threshold (such as 50%) may be specifically determined as target drivers. In practical applications, the target driver to be called can also be determined by combining the traffic gap or the driver demand of the area and the awakening probability of each silent driver.

Step S712, obtaining the short message contact information of the target driver, and sending the dispatching notice reaching the area in the appointed time period to the target driver in the form of short message. To dispatch the target driver to the area of insufficient capacity.

The identification of the area with insufficient capacity and the identification of the designated period can be included in the scheduling notice so that the driver can clearly know the scheduling area and the time; of course, in order to increase the driver's motivation and prompt the driver to coordinate with the dispatch, the dispatch notice may further include an incentive mechanism, which may include a coupon for an order in the response area and/or an order drawing ratio, etc.

Through the above mode that this embodiment provided, can predict the not enough region of transport capacity, and find out the silence driver that this region is relevant, through the awakening probability of each silence driver that obtains of calculation, select a batch of target driver from the silence driver, and then adopt to send the SMS to it, awaken it for its mode such as reward, guide it to go to corresponding region and receive the order, effectively alleviate passenger's the vigorous problem of driver supply but the driver is not enough of passenger's car demand, not only reduced passenger's the time spent of taking the car, but also brought the income for the driver, increased passenger and driver to the adhesion degree of platform of taking the car, user experience has been promoted betterly.

Example four

Corresponding to the foregoing capacity scheduling method, the embodiment provides a capacity scheduling apparatus, which is applied to a platform server, and referring to a block diagram of a structure of the capacity scheduling apparatus shown in fig. 8, the capacity scheduling apparatus includes a driver determining module 82, a driver screening module 84, and a driver scheduling module 86, which are connected in sequence, wherein:

a driver determination module 82, configured to determine data including silent driver information according to historical driver data corresponding to an area if an area with insufficient transportation capacity is predicted within a specified time period, where the silent driver includes an offline driver with an active value within a set active value interval; .

In one embodiment, the driver determination module is configured to: searching candidate drivers in the area according to the historical driver data; the historical driver data includes one or more of the following information: a history active area, a registered home address, a vehicle receiving place and a vehicle leaving place; and determining the silent driver according to the online records of the candidate drivers in a specified time period.

In particular implementations, the driver determination module can be configured to find candidate drivers according to the following steps: performing GeoHash coding on the region according to longitude and latitude coordinates; acquiring an inverted index value corresponding to the region according to the GeoHash code corresponding to the region; determining data containing driver information corresponding to the inverted index value in an inverted index table as data containing candidate driver information corresponding to the region; the reverse index table is a corresponding relation table of reverse index values and driver information which is established in advance according to historical driver data.

In another embodiment, the driver determination module is configured to: searching for an online record of data containing candidate driver information in a specified time period, wherein the online record comprises: receiving and/or leaving records; and filtering online drivers in the candidate drivers and offline drivers with the activity values lower than the lower limit value of the set activity value interval to obtain silent drivers.

A driver screening module 84 for screening data containing target driver information from data containing silent driver information.

In one embodiment, the driver screening module is configured to: screening data containing target driver information from the data containing silent driver information according to the insufficient degree of the transport capacity of the area; or screening the data containing the target driver information from the data containing the silent driver information according to the awakening probability of the silent driver.

In particular implementations, the driver screening module can be configured to determine the target driver according to the following steps: determining the transport capacity gap number of the area according to the average value of unanswered orders of the area in a specified time period; counting the number of silent drivers; determining the number of target drivers according to the number of silent drivers and the number of capacity gaps; and selecting the data containing the driver information of the quantity as the data containing the target driver information from the data containing the silent driver information.

In another embodiment, the driver screening module is configured to: calculating the awakening probability of the silent driver based on a pre-trained machine learning model; and determining the data containing the silent driver information, of which the awakening probability is greater than a set second threshold value, as the data containing the target driver information.

Wherein, the training process of the machine learning model comprises the following steps: training a preset XGboost model according to a historical awakening database of a silent driver; acquiring feature data output by the trained XGboost model, and inputting the feature data into a preset LR model for training; and taking the trained LR model as a machine learning model.

A driver scheduling module 86 for issuing a scheduling notice to the target driver to reach the above-mentioned area (i.e., the aforementioned area with insufficient capacity) within a specified period of time.

In one embodiment, the driver scheduling module is to: acquiring a short message contact way of a target driver; and issuing a scheduling notice reaching the region within a specified time period to the target driver by using a short message contact way, wherein the scheduling notice comprises the identifier of the region and the identifier of the specified time period.

In another embodiment, the driver scheduling module is to: acquiring a reward mechanism and scheduling information of a region; wherein, the reward mechanism includes: coupons and/or order draws to answer the order in the area; the scheduling information includes: an identification of the region and an identification of the designated time period; and issuing a scheduling notice reaching the area within a specified time period to the target driver, wherein the scheduling notice comprises a reward mechanism and scheduling information.

The capacity scheduling device provided by the embodiment of the application can determine data containing silent driver information according to historical driver data corresponding to a capacity shortage area, screen data containing target driver information from the data containing silent driver information, and then send a scheduling notice to a target driver, wherein the scheduling notice reaches the capacity shortage area within a specified time period. This kind of select the target driver that can dispatch from silent driver to dispatch target driver and go to the regional mode that the capacity is not enough, can alleviate passenger better and use the vigorous but insufficient supply of driver's demand problem of car, effectively promoted user experience.

Referring to a block diagram of another capacity scheduling apparatus shown in fig. 9, on the basis of fig. 8, the apparatus further includes: and the area prediction module 92 is used for predicting whether an area with insufficient transport capacity exists in a specified time period according to historical order data.

In one embodiment, the region prediction module is configured to: searching a hot spot area corresponding to a specified time period; calling at least one group of historical order data of the hot spot region corresponding to the designated time period; counting the average value of unanswered orders in each group of historical order data; and if the average value exceeds a set first threshold value, determining the hot spot area as an area with insufficient transport capacity corresponding to the designated period.

Further, the area prediction module is specifically configured to obtain order data according to the following steps: taking the current time as a reference, calling historical order data of N designated time periods corresponding to the hot spot region to obtain N groups of historical order data; wherein N is a preset natural number.

The device provided by the embodiment has the same implementation principle and technical effect as the foregoing embodiment, and for the sake of brief description, reference may be made to the corresponding contents in the foregoing method embodiment for the portion of the embodiment of the device that is not mentioned.

EXAMPLE five

An embodiment of the present application provides a server, which includes a memory and a processor, wherein the memory is used for storing a program for supporting the processor to execute any one of the foregoing capacity scheduling methods, and the processor is configured to execute the program stored in the memory.

Referring to the schematic structural diagram of a server shown in fig. 10, specifically, the server includes a processor 100, a memory 101, a bus 102 and a communication interface 103, where the processor 100, the communication interface 103 and the memory 101 are connected through the bus 102; the processor 100 is adapted to execute executable modules, such as computer programs, stored in the memory 101.

The Memory 101 may include a high-speed Random Access Memory (RAM) and may also include a non-volatile Memory (non-volatile Memory), such as at least one disk Memory. The communication connection between the network element of the system and at least one other network element is realized through at least one communication interface 103 (which may be wired or wireless), and the internet, a wide area network, a local network, a metropolitan area network, and the like can be used.

The bus 102 may be an ISA bus, PCI bus, EISA bus, or the like. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of illustration, only one double-headed arrow is shown in FIG. 10, but this does not indicate only one bus or one type of bus.

The memory 101 is used for storing a program, the processor 100 executes the program after receiving an execution instruction, and the method performed by the apparatus defined by the flow program disclosed in any of the foregoing embodiments of the present application may be applied to the processor 100, or implemented by the processor 100.

Processor 100 may be an integrated circuit chip having signal processing capabilities. In implementation, the steps of the above method may be performed by integrated logic circuits of hardware or instructions in the form of software in the processor 100. The Processor 100 may be a general-purpose Processor, and includes a Central Processing Unit (CPU), a Network Processor (NP), and the like; the device can also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA) or other Programmable logic device, a discrete Gate or transistor logic device, or a discrete hardware component. The various methods, steps, and logic blocks disclosed in the embodiments of the present application may be implemented or performed. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like. The steps of the method disclosed in connection with the embodiments of the present application may be directly implemented by a hardware decoding processor, or implemented by a combination of hardware and software modules in the decoding processor. The software module may be located in ram, flash memory, rom, prom, or eprom, registers, etc. storage media as is well known in the art. The storage medium is located in the memory 101, and the processor 100 reads the information in the memory 101 and completes the steps of the method in combination with the hardware.

The transportation capacity scheduling method provided in this embodiment may be executed by the server, or the transportation capacity scheduling method apparatus provided in this embodiment may be disposed on the server side.

Further, the present embodiment also provides a computer storage medium for storing computer software instructions for any one of the capacity scheduling devices.

Specifically, the storage medium can be a general storage medium, such as a mobile disk, a hard disk, and the like, and when a computer program on the storage medium is run, the capacity scheduling method can be executed, so that the problem that in the prior art, a passenger is difficult to call a vehicle due to insufficient supply and demand of a driver is solved, and user experience is effectively improved.

The functions, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present application or portions thereof that substantially contribute to the prior art may be embodied in the form of a software product stored in a storage medium and including instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present application. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.

In the description of the present application, it is noted that the terms "first", "second", and "third" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.

Finally, it should be noted that: the above-mentioned embodiments are only specific embodiments of the present application, and are used for illustrating the technical solutions of the present application, but not limiting the same, and the scope of the present application is not limited thereto, and although the present application is described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: any person skilled in the art can modify or easily conceive the technical solutions described in the foregoing embodiments or equivalent substitutes for some technical features within the technical scope disclosed in the present application; such modifications, changes or substitutions do not depart from the spirit and scope of the exemplary embodiments of the present application, and are intended to be covered by the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims

1. A capacity scheduling method is applied to a platform server, and comprises the following steps:

if an area with insufficient transport capacity in a specified time period is predicted, determining data containing silent driver information according to historical driver data corresponding to the area; wherein the silent drivers comprise off-line drivers with an active value within a set active value interval;

screening data containing target driver information from the data containing the silent driver information;

issuing a dispatch notice to the target driver to reach the zone within the specified time period.

2. The method of claim 1, wherein the method further comprises:

and predicting whether an area with insufficient transport capacity exists in a specified time period according to historical order data.

3. The method of claim 2, wherein the step of predicting from historical order data whether an area of insufficient capacity exists within a specified time period comprises:

searching a hot spot area corresponding to a specified time period;

calling at least one group of historical order data of the hot spot region corresponding to the specified time period;

counting the average value of unanswered orders in each group of the called historical order data;

and if the average value exceeds a set first threshold value, determining the hot spot area as an area with insufficient transport capacity corresponding to the specified time period.

4. The method of claim 3, wherein said step of retrieving at least one set of historical order data for said hotspot region corresponding to said specified time period comprises:

taking the current time as a reference, calling N sets of historical order data of the appointed time periods corresponding to the hot spot region to obtain N sets of historical order data; wherein N is a preset natural number.

5. The method of claim 1, wherein the step of determining data containing silent driver information based on historical driver data corresponding to the region comprises:

searching data containing candidate driver information of the area according to historical driver data; the historical driver data includes one or more of the following information: a history active area, a registered home address, a vehicle receiving place and a vehicle leaving place;

and determining data containing silent driver information according to the online record of the data containing the candidate driver information in the specified time period.

6. The method of claim 5, wherein the step of searching for data containing candidate driver information for the area based on historical driver data comprises:

performing GeoHash coding on the region according to longitude and latitude coordinates;

acquiring an inverted index value corresponding to the region according to the GeoHash code corresponding to the region;

determining data containing driver information corresponding to the inverted index value in an inverted index table as data containing candidate driver information corresponding to the region; the reverse index table is a corresponding relation table of reverse index values and driver information which is established in advance according to historical driver data.

7. The method of claim 5, wherein the step of determining data containing silent driver information based on online recording of data containing the candidate driver information over the specified time period comprises:

finding an online record of data containing the candidate driver information over the specified time period, the online record comprising: receiving and/or leaving records;

and filtering the data containing the online driver information and the data containing the offline driver information, the activity value of which is lower than the lower limit value of the set activity value interval, in the data containing the candidate driver information to obtain the data containing the silent driver information.

8. The method of claim 1, wherein the step of filtering the data containing target driver information from the data containing silent driver information comprises:

screening data containing target driver information from the data containing silent driver information according to the insufficient degree of the transport capacity of the area; alternatively, the first and second electrodes may be,

and screening data containing target driver information from the data containing the silent driver information according to the awakening probability of the silent driver.

9. The method of claim 8, wherein the step of screening the data containing the driver information of the target from the data containing the driver information of the silent driver according to the degree of the insufficient capacity of the area comprises:

determining the transport capacity gap number of the area according to the average value of unanswered orders of the area in the specified time period;

counting the number of silent drivers;

determining the number of target drivers according to the number of silent drivers and the number of the capacity gaps;

and selecting the data containing the number of driver information from the data containing the silent driver information as the data containing the target driver information.

10. The method of claim 8, wherein the step of filtering the data containing the target driver information from the data containing the silent driver information according to the awakening probability of the silent driver comprises:

calculating a wake-up probability of the silent driver based on a pre-trained machine learning model;

and determining the data containing the silent driver information, of which the awakening probability is greater than a set second threshold value, as the data containing the target driver information.

11. The method of claim 10, wherein the training process of the machine learning model comprises:

training a preset Xgboost model according to a historical awakening database of the silent driver;

acquiring feature data output by the trained Xgboost model, and inputting the feature data into a preset LR model for training;

and taking the trained LR model as the machine learning model.

12. The method of claim 1, wherein the step of issuing a dispatch notification to a target driver to reach the zone within the specified time period comprises:

acquiring a short message contact way of the target driver;

and issuing a scheduling notice reaching the area within the specified time period to the target driver by using the short message contact way, wherein the scheduling notice comprises the identification of the area and the identification of the specified time period.

13. The method of claim 1, wherein the step of issuing a dispatch notification to a target driver to reach the zone within the specified time period comprises:

acquiring a reward mechanism and scheduling information of the area; wherein the reward mechanism comprises: coupons and/or order draws that answer orders for the area are proportioned; the scheduling information includes: an identification of the region and an identification of the specified time period;

issuing a dispatch notification to the target driver to reach the zone within the specified time period, the dispatch notification including the reward mechanism and the dispatch information.

14. A capacity scheduling apparatus applied to a platform server, the apparatus comprising:

the driver determining module is used for determining data containing silent driver information according to historical driver data corresponding to an area if an area with insufficient transport capacity in a specified time period is predicted; wherein the silent drivers comprise off-line drivers with an active value within a set active value interval;

the driver screening module is used for screening data containing target driver information from the data containing the silent driver information;

and the driver scheduling module is used for issuing a scheduling notice reaching the area within the specified time period to the target driver.

15. The apparatus of claim 14, wherein the apparatus further comprises:

and the area prediction module is used for predicting whether an area with insufficient transport capacity exists in a specified time period according to historical order data.

16. The apparatus of claim 15, wherein the region prediction module is to:

searching a hot spot area corresponding to a specified time period;

17. The apparatus of claim 16, wherein the region prediction module is to:

18. The apparatus of claim 14, wherein the driver determination module is to:

19. The apparatus of claim 18, wherein the driver determination module is to:

20. The apparatus of claim 18, wherein the driver determination module is to:

21. The apparatus of claim 14, wherein the driver screening module is to:

22. The apparatus of claim 21, wherein the driver screening module is to:

counting the number of silent drivers;

23. The apparatus of claim 21, wherein the driver screening module is to:

24. The apparatus of claim 23, wherein the training process of the machine learning model comprises:

and taking the trained LR model as the machine learning model.

25. The apparatus of claim 14, wherein the driver scheduling module is to:

acquiring a short message contact way of the target driver;

26. The apparatus of claim 14, wherein the driver scheduling module is to:

27. A server, characterized in that the server comprises a memory for storing a program enabling a processor to perform the method of any of claims 1 to 13 and a processor configured for executing the program stored in the memory.

28. A computer storage medium storing computer software instructions for use by the apparatus of any one of claims 14 to 26.