CN112435462B - Method, system, electronic device and storage medium for short-time traffic flow prediction - Google Patents
Method, system, electronic device and storage medium for short-time traffic flow prediction Download PDFInfo
- Publication number
- CN112435462B CN112435462B CN202011111809.3A CN202011111809A CN112435462B CN 112435462 B CN112435462 B CN 112435462B CN 202011111809 A CN202011111809 A CN 202011111809A CN 112435462 B CN112435462 B CN 112435462B
- Authority
- CN
- China
- Prior art keywords
- data
- features
- road
- prediction
- time
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
- G08G1/0125—Traffic data processing
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/065—Traffic control systems for road vehicles by counting the vehicles in a section of the road or in a parking area, i.e. comparing incoming count with outgoing count
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Analytical Chemistry (AREA)
- Chemical & Material Sciences (AREA)
- Theoretical Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- General Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Traffic Control Systems (AREA)
Abstract
The application relates to a method, a system, an electronic device and a storage medium for short-time traffic flow prediction, wherein the short-time traffic flow prediction method comprises the following steps: acquiring multi-modal road network source data, preprocessing the data by taking road sections and time as main dimensions, and generating basic data with uniform caliber and matched road sections; respectively extracting features of road network spatio-temporal data, environmental data and video image data in the basic data to obtain spatio-temporal dependence features, environmental features and image features; performing feature fusion on the obtained features to obtain fusion features; filtering the fusion features through an attention mechanism model, and capturing key effective features in the fusion features; and finally, inputting the effective characteristics into a regression model for prediction to obtain a predicted value. By the method and the device, the problems of low utilization rate of multi-mode data of the road network and low accuracy rate of multi-condition prediction of road conditions are solved, and the accuracy and the efficiency of model prediction are improved.
Description
Technical Field
The present application relates to the field of artificial intelligence, and more particularly, to methods, systems, electronic devices, and storage media for short-term traffic flow prediction.
Background
With the continuous development and construction of the expressway, more and more people have been used as the main choice of daily travel mode due to the high-efficiency and rapid traffic capacity of the expressway. However, with the increasing travel demand of people, the traffic capacity of the existing road network can not meet the existing travel amount, and more traffic jam conditions are generated immediately, and the consequence is the frequent occurrence of traffic accidents. Therefore, the prediction of the short-time traffic volume of the expressway is a necessary premise for realizing reasonable and effective traffic induction, relieving traffic jam and reducing the frequent occurrence of traffic accidents.
In the related technology, linear regression-based algorithm prediction is to obtain factors possibly influencing traffic flow to carry out feature scaling, initialize parameters of the factors subjected to feature scaling, model the factors subjected to feature scaling by combining a linear regression method, determine a cost function according to actual traffic flow and an established model, and carry out regularization processing on the cost function; and then solving the minimum value of the cost function by using a gradient descent algorithm, solving an optimal parameter according to the minimum value of the cost function, and finally predicting the short-term traffic flow of the road by using the solved optimal parameter. The linear regression-based algorithm prediction assumes that the acquired road network data is in a linear trend, the output data is only in a linear relation, the road network data is diversified, and the flow of the road network is also randomly changed, so the linear regression algorithm has large error and low data utilization rate, and the emergency can not be identified;
the method based on the convolutional neural network mainly utilizes the construction of a traffic flow sequence, inputs the flow sequence into the convolutional neural network for training so as to obtain the prediction result of an observation point, and the main algorithm steps are as follows: (1) selecting road sections needing traffic flow prediction and available corresponding road network historical data; (2) selecting a prediction time period for predicting the short-term traffic flow according to the acquired historical data of the short-term traffic flow; (3) performing two-dimensional matrix transformation on historical traffic flow data of the predicted road section; (4) and constructing a convolutional neural network, carrying out prediction model training on the historical traffic flow data set, and verifying and optimizing. The input of prediction data is a single source, the problem of road network multi-modal data application is not solved, in addition, the convolutional neural network can only complete the prediction of a single task, the simultaneous prediction of various road conditions cannot be realized, the operation efficiency is low, and the calculation cost is high;
the method comprises the following steps of predicting, based on a support vector machine supervised training model, data by using different kernel functions to carry out transformation processing, mapping nonlinearly to a high latitude feature space, constructing an optimal classification hyperplane in the high dimensional feature space, and finally fitting or classifying by using the hyperplane, wherein the main algorithm steps are as follows: (1) collecting original traffic flow data passing through a fixed road at fixed time intervals; (2) preprocessing original traffic flow data according to a seasonal model algorithm to generate time sequence traffic flow data; (3) selecting a proper kernel function according to the data type and the data dimension, and constructing a regression model based on a support vector machine; (4) and inputting the time sequence data into a support vector machine model for training and testing, and adjusting model parameters according to the average absolute percentage error of the prediction result. The prediction accuracy of the prediction method for large sample size is reduced, multi-modal data cannot be processed, the spatial dependence of the road network cannot be captured by the extracted features, and the complexity and randomness of the road network cannot be reflected by the features participating in model calculation.
At present, no effective solution is provided for the problems of low utilization rate of multi-mode data of a road network and low accuracy rate of multi-mode road condition prediction under the condition of predicting the road condition of a short distance in the middle of an expressway in the related technology.
Disclosure of Invention
The embodiment of the application provides a method, a system, an electronic device and a storage medium for predicting short-time traffic flow, and at least solves the problems that in the related art, under the condition of predicting road conditions in a short distance in the middle of an expressway, the utilization rate of multi-mode data of a road network is low, and the accuracy rate of predicting the road conditions under multiple conditions is low.
In a first aspect, an embodiment of the present application provides a method for short-time traffic flow prediction, where the method includes:
acquiring multi-modal road network source data, preprocessing the data by taking a road section and time as main dimensions, and generating basic data with uniform caliber and matched road sections;
respectively extracting the characteristics of the road network space-time data, the environmental data and the video image data in the basic data to obtain time dependence characteristics, space dependence characteristics, environmental characteristics and image characteristics;
performing feature fusion on the time-dependent features, the space-dependent features, the environmental features and the image features to obtain fusion features;
filtering the fusion features through an attention mechanism model, and capturing key effective features in the fusion features;
and inputting the effective characteristics into a task regression model for prediction to obtain a predicted value.
In some of these embodiments, the filtering the fused features by the attention mechanism model comprises:
establishing attention weight among the fusion features to generate an attention weight matrix;
performing softmax normalization on the attention weight matrix to obtain a weight vector;
and carrying out weighted linear combination on the fusion features through the weight vector to obtain a feature sequence with weight reference.
In some embodiments, the feature extracting road network spatiotemporal data, environmental data and video image data in the road network source data respectively comprises:
and extracting the features of the road network space-time data according to a T-GCN model to obtain time-dependent features and space-dependent features, extracting the environmental data by adopting one-hot features to obtain environmental features, and extracting the video image data by adopting convolutional neural network features to obtain image features.
In some embodiments, the generating the base data of uniform caliber and matched road sections comprises:
defining a calculation caliber, and converting the road network source data to obtain basic data with uniform caliber;
and splitting or aggregating the road network source data based on road sections to obtain basic data matched with the road sections.
In some embodiments, the inputting the new feature into the task regression model for prediction includes:
and simultaneously predicting a plurality of traffic flow data of the specified road section of the specified time window in a multitask mode.
In a second aspect, an embodiment of the present application provides a system for short-time traffic flow prediction, where the system includes:
the acquisition and preprocessing module is used for acquiring multi-modal road network source data, preprocessing the data by taking a road section and time as main dimensions, and generating basic data with uniform caliber and matched road sections;
the characteristic extraction module is used for respectively extracting the characteristics of the road network space-time data, the environmental data and the video image data in the basic data to obtain time dependence characteristics, space dependence characteristics, environmental characteristics and image characteristics;
the characteristic fusion module is used for carrying out characteristic fusion on the time-dependent characteristic, the space-dependent characteristic, the environment characteristic and the image characteristic to obtain a fusion characteristic;
the attention mechanism module is used for filtering the fusion features through an attention mechanism model and capturing key effective features in the fusion features;
and the traffic flow prediction module is used for inputting the effective characteristics into the task regression model for prediction to obtain a predicted value.
In some of these embodiments, the filtering the fused features by the attention mechanism model comprises:
the attention mechanism module establishes attention weights among the fusion features to generate an attention weight matrix;
performing softmax normalization on the attention weight matrix to obtain a weight vector;
and carrying out weighted linear combination on the fusion features through the weight vector to obtain a feature sequence with weight reference.
In a third aspect, an embodiment of the present application provides an electronic device, including a memory and a processor, where the memory stores a computer program, and the processor is configured to execute the computer program to perform the method for short-time traffic flow prediction according to any one of the foregoing methods.
In a fourth aspect, the present application provides a storage medium, in which a computer program is stored, where the computer program is configured to execute the short-time traffic flow prediction method according to any one of the above methods when the computer program is executed.
Compared with the related technology, the method for predicting the short-time traffic flow, provided by the embodiment of the application, comprises the steps of obtaining multi-mode road network source data, preprocessing the data by taking a road section and time as main dimensions, and generating basic data with uniform caliber and matched road sections; respectively extracting the characteristics of road network space-time data, environmental data and video image data in the basic data to obtain time dependence characteristics, space dependence characteristics, environmental characteristics and image characteristics; performing feature fusion on the time-dependent feature, the space-dependent feature, the environmental feature and the image feature to obtain a fusion feature; filtering the fusion features through an attention mechanism model, and capturing key effective features in the fusion features; and finally, the obtained effective characteristics are input into the task regression model for prediction to obtain predicted values, and the problems of low utilization rate of multi-mode data of a road network and low accuracy rate of multi-mode prediction of the road conditions under the condition of predicting the road conditions in a short distance in the middle of an expressway are solved, so that the short-time traffic flow prediction is more accurate, the outburst of the short-distance road conditions is better predicted, and the accuracy and the efficiency of model prediction are improved.
Drawings
The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this application, illustrate embodiment(s) of the application and together with the description serve to explain the application and not to limit the application. In the drawings:
FIG. 1 is a schematic diagram of an application environment of a method for short-term traffic flow prediction according to an embodiment of the application;
FIG. 2 is a flow chart of a short-term traffic flow prediction method according to an embodiment of the present application;
FIG. 3 is a block diagram of a short-time traffic flow prediction system according to an embodiment of the present application;
FIG. 4 is a data flow diagram illustrating a short-term traffic flow prediction according to an embodiment of the present application;
fig. 5 is an internal structural diagram of an electronic device according to an embodiment of the present application.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application will be described and illustrated below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments provided in the present application without any inventive step are within the scope of protection of the present application. Moreover, it should be appreciated that in the development of any such actual implementation, as in any engineering or design project, numerous implementation-specific decisions must be made to achieve the developers' specific goals, such as compliance with system-related and business-related constraints, which may vary from one implementation to another.
Reference in the specification to "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the specification. The appearances of the phrase in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments. Those of ordinary skill in the art will explicitly and implicitly appreciate that the embodiments described herein may be combined with other embodiments without conflict.
Unless defined otherwise, technical or scientific terms referred to herein shall have the ordinary meaning as understood by those of ordinary skill in the art to which this application belongs. Reference to "a," "an," "the," and similar words throughout this application are not to be construed as limiting in number, and may refer to the singular or the plural. The present application is directed to the use of the terms "including," "comprising," "having," and any variations thereof, which are intended to cover non-exclusive inclusions; for example, a process, method, system, article, or apparatus that comprises a list of steps or modules (elements) is not limited to the listed steps or elements, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus. Reference to "connected," "coupled," and the like in this application is not intended to be limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. Reference herein to "a plurality" means greater than or equal to two. "and/or" describes an association relationship of associated objects, meaning that three relationships may exist, for example, "A and/or B" may mean: a exists alone, A and B exist simultaneously, and B exists alone. Reference herein to the terms "first," "second," "third," and the like, are merely to distinguish similar objects and do not denote a particular ordering for the objects.
The short-term traffic flow prediction method provided by the application can be applied to an application environment shown in fig. 1, fig. 1 is an application environment schematic diagram of the short-term traffic flow prediction method according to the embodiment of the application, and as shown in fig. 1, a system of the application environment includes a server 10 and a high-speed management terminal device 11, where the server 10 acquires multi-modal road network source data, and generates basic data with uniform caliber and matched road sections by preprocessing the data with road sections and time as major dimensions; respectively extracting the features of road network space-time data, environmental data and video image data in the preprocessed basic data to obtain time dependence features, space dependence features, environmental features and image features; then, performing feature fusion on the acquired time dependence features, space dependence features, environment features and image features to obtain fusion features; filtering the fusion features through an attention mechanism model, and capturing key effective features in the fusion features; and finally, inputting the obtained effective characteristics into a task regression model for prediction to obtain a predicted value, and sending the predicted value to the high-speed management terminal device 11, so that the problems of low utilization rate of multi-mode data of a road network and low accuracy rate of multi-situation prediction of road conditions under the condition of predicting the road conditions in a short distance in the middle of an expressway are solved, the short-time traffic flow prediction is more accurate, the outburst of the short-distance road conditions is better predicted, the high-speed management terminal device 11 can judge the traffic road conditions of the corresponding road sections in the future according to the traffic flow at the future time, and preventive scheduling measures are taken for the road sections which are likely to generate congestion.
The embodiment provides a method for predicting short-term traffic flow, fig. 2 is a flowchart of a method for predicting short-term traffic flow according to an embodiment of the present application, and as shown in fig. 2, the flowchart includes the following steps:
step S201, obtaining multi-mode road network source data, obtaining required related field data of three modes from the source data based on a predicted target road section, preprocessing the data by taking the road section and time as main dimensions, and generating basic data with uniform caliber and matched road sections, wherein the multi-mode is multiple information sources and forms, in the embodiment, part of the multi-mode road network source data is derived from vehicle traffic data and image stream data which are collected by an ETC portal equipment system laid on an expressway, and the data covers license plates, vehicle speeds, lanes and shooting time information of passing vehicles; the other part is derived from static data tables and third party interface data provided by the high speed management company, which are mainly related to road infrastructure and weather. Optionally, the field data related to the three modalities respectively include: the first type is the traffic data in an ETC portal system, and the pure structured data is the most important basic data source in the model and mainly comprises the information of serial numbers, ETC portal road network numbers, portal road section numbers, ETC portal frame numbers, vehicle types, license plate numbers, license plate colors, passing time, driving directions, vehicle types and the like. The second type is image flow recording data, which is used for supplementing flow data on one hand and providing accident data and behavior data such as dangerous driving of an expressway on the other hand. The image flow data mainly records the flow number, the serial number of an ETC road network, the serial number of an ETC portal frame, equipment codes, road codes, license plate numbers, license plate colors, vehicle speeds and the vehicle running direction. The third type is the secondary processing of external third party data and static data set by road foundation, the third party data mainly includes weather data and geographical position data, the secondary processing of the basic data includes whether the road section contains junction, road width and number of lanes, whether the road section is far away from city, whether the road section is close to toll station port, average speed limit of the road section, etc. In this embodiment, the acquired road network data is preprocessed to generate basic data with uniform caliber and matched road sections, and optionally, the preprocessing is mainly divided into two parts: the first part is data cleaning, and because different data sources are from different devices and acquisition platforms, the problems of data loss, recording errors, repeated data and the like caused by device failure or insufficient device performance are solved, so that the error data and abnormal data need to be corrected, missing road data is made up, and redundant data is removed; the second part is to carry out uniform caliber and road section matching on data collected by different devices. The embodiment aims at the complexity and the changeability of road network data, the data are divided into three modes for processing, the problem of low utilization rate of multi-mode data of the road network is solved, the utilization rate of the data is improved, and the method is favorable for predicting and identifying different conditions of roads in the follow-up process.
Step S202, respectively extracting the characteristics of the road network spatio-temporal data, the environmental data and the video image data in the preprocessed basic data to obtain time dependence characteristics, space dependence characteristics, environmental characteristics and image characteristics, wherein the space dependence refers to flow change caused by the influence of the topological structure of the road network, for example, the traffic state of an upstream highway influences a downstream highway through transfer, and the traffic state of the downstream highway influences the upstream through feedback; time-dependent means that the traffic flow dynamically changes with time, and a periodic trend occurs, and the highway traffic flow is influenced by the traffic state at the previous moment or the later moment. For example, data such as ETC portal road network numbers, ETC portal road section numbers, license plate colors, snapshot time and the like generated by a high-speed ETC portal system are sorted and mined, regression and prediction are performed on information such as traffic flow and vehicle speed by using time sequence fitting, historical data such as special dates and holidays are analyzed, specific periodic events or trends are extracted, event influences between upstream and downstream expressways are effectively predicted, and whether special conditions occur is deduced. Compared with the prior art that the input of the feature of the predicted data is single, the prediction of a single task can only be completed, and the feature participating in model calculation cannot reflect the complexity and randomness of the road network.
Step S203, performing feature fusion on the acquired time-dependent feature, space-dependent feature, environmental feature and image feature to obtain a fusion feature, in this embodiment, the time-dependent feature, the space-dependent feature, the environmental feature and the image feature are serially fused in a single-layer full-connection manner by using a neural network method, and information of each type of feature is retained.
Step S204, filtering the obtained fusion features through an attention mechanism model, and capturing key effective features in the fusion features, wherein an attention mechanism (attention mechanism) is a resource allocation scheme of a main means for solving the information overload problem, and allocates computing resources to more important tasks, and the attention is generally divided into two types: one is conscious attention from top to bottom, called focused attention, which refers to attention that has a predetermined purpose, is task-dependent, and is actively and consciously focused on a certain subject; the other is involuntary attention from bottom to top, called saliency-based attention, which is attention driven by external stimuli, does not require active intervention, and is also task independent. In the embodiment, an attention mechanism model is added after multi-modal feature fusion, the fusion features are filtered according to predicted road section targets, and key effective features in the fusion features are captured, and because the feature fusion is not really completed after single-layer full-connection splicing and fusion are adopted for various modal features, the fusion features are also required to be filtered and captured by the attention mechanism, so that effective space dependence features of special road section information, effective time dependence features of special time section information, effective environment features and effective image features are obtained, especially the effective space dependence feature capture of the special road section information and the effective time dependence feature capture of the special time section information are increased, and optionally, the special road section information refers to road section information such as abnormal road sections, accident-prone road sections, confluence parts and the like; optionally, the special time period information refers to time information such as morning and evening peaks, holidays, seasonal regularity information and the like, compared with the related art, processed features are directly used as the input parameters to enter the model for prediction, the key features obtained through filtering are used as the input parameters to enter the model for prediction through the attention mechanism model, the real situation of high-speed short-time traffic flow can be reflected, the problem of low accuracy of multi-situation road condition prediction is solved, and the prediction accuracy of the model is improved.
And S205, inputting the obtained effective characteristics into a task regression model for prediction to obtain a final predicted value. The task regression model refers to a predictive mathematical model, and mainly studies the relationship between a target and a predictor, for example, the predictive analysis of high-speed short-time traffic flow. In the embodiment, the obtained effective characteristics are input into the task regression model for prediction to obtain the predicted value, the predicted value is sent to the high-speed management terminal device 11, the high-speed management operation company is helped to judge whether the corresponding road section in the future is smooth or not according to the predicted value of the traffic flow, and preventive scheduling measures are taken for the road section which is likely to cause congestion, so that the problem of predicting the sudden situation of the short-distance complex road condition is solved, and the accuracy and fineness of prediction are improved.
Through the steps S201 to S205, compared to the prior art, the algorithm based on linear regression has large prediction error, low data utilization rate, and cannot recognize an emergency, the method based on the convolutional neural network can only complete the prediction of a single task, cannot realize the simultaneous prediction of multiple road conditions, has low operation efficiency and high calculation cost, and the prediction accuracy of a large sample size based on the support vector machine supervised training model prediction is reduced, cannot process multi-modal data, and cannot reflect the problems of complexity and randomness of the road network. In the method based on the multi-modal attention mechanism model, the server 10 acquires multi-modal road network source data, preprocesses the data by taking road sections and time as main dimensions, and generates basic data with uniform caliber and matched road sections; respectively extracting the features of road network space-time data, environmental data and video image data in the preprocessed basic data to obtain time dependence features, space dependence features, environmental features and image features; then, performing feature fusion on the acquired time dependence features, space dependence features, environment features and image features to obtain fusion features; filtering the fusion features through an attention mechanism model, and capturing key effective features in the fusion features; and finally, inputting the obtained effective characteristics into the task regression model for prediction to obtain a predicted value, and sending the predicted value to the high-speed management terminal device 11, so that the problems of low utilization rate of multi-mode data of a road network and low accuracy rate of multi-situation prediction of road conditions under the condition of predicting the road conditions in a short distance in the middle of a highway are solved, the short-time traffic flow prediction is more accurate, the outburst of the short-distance road conditions is better predicted, and the accuracy and the efficiency of model prediction are improved.
In some of these embodiments, filtering the fused features through the attention mechanism model includes: establishing attention weight among the fusion features, generating an attention weight matrix, carrying out softmax normalization on the attention weight matrix to obtain a weight vector, carrying out weighted linear combination on the fusion features through the weight vector to obtain a feature sequence with a weight reference, wherein the feature sequence comprises: the method comprises the steps of adopting a space attention mechanism to capture space characteristic sequences of special road sections such as abnormal road sections, accident-prone road sections and confluence parts and adopting a time attention mechanism to capture time characteristic sequences of special time periods such as morning and evening peaks, holidays and seasonal regularity information. In the embodiment, the fusion features are filtered through the attention mechanism model to obtain the key effective time features and space features, compared with the prior art center, the processed features are directly used as the input parameters to enter the model prediction, the embodiment can better reflect the real situation of the high-speed short-time traffic flow, the problem of low accuracy rate of the multi-situation road condition prediction is solved, and the accuracy of the model prediction is improved.
In some embodiments, the feature extraction of the road network spatiotemporal data, the environmental data and the video image data in the road network source data respectively comprises: performing feature extraction on road network space-time data according to a T-GCN model to obtain time dependence features and space dependence features, performing one-hot feature extraction on the environment data to obtain environment features, performing convolutional neural network feature extraction on video image data to obtain image features, in the embodiment, time sequence traffic information of a road section at the current time T is obtained by matching relations among the road sections, taking historical time sequence data with the length of n as input, receiving space information of a topological structure by using the GCN, inputting the received space and time information into a GRU, and obtaining dynamic information change among units to extract the time dependence features, and capturing the space dependence features among nodes through a first-order neighborhood of the GCN model for a given road section adjacent matrix and feature matrix; optionally, in this embodiment, the environment data includes road condition data and environment data, where the road condition data mainly refers to a geographic location attribute of a road segment, for example, whether the road segment is close to a toll station, whether the road segment includes a pivot point, and the number of entrances/exits of a road segment node, and the environment data mainly refers to a weather condition, including current weather and future predicted weather conditions, and such characteristic dimensions are not high, and one-hot can be directly used for feature extraction to obtain environmental features, where one-hot is also called "one-hot coding", which is a method for more commonly used text feature extraction, and N states are encoded by using N-bit state registers, each state has an independent register bit, and only one of the register bits is valid; optionally, in this embodiment, the video image data mainly includes abnormal road section feature data and lane data, and data acquired by the ETC portal device does not relate to data at a lane level, and there is no data of road section construction, accidents, vehicle road occupation ratio, and the like, so that feature supplementation for road conditions can be obtained by analyzing vehicle distances at the lane level and learning other image features by using the video image data.
In some of these embodiments, generating the aperture unification and segment matching base data comprises: defining a calculation caliber, and converting road network source data to obtain basic data with uniform calibers; the road network source data is split or aggregated, and corresponding to the small road segments, the basic data matched with the road segments is obtained, and optionally, because the data acquired by different devices has the problem of inconsistent aperture, the embodiment converts the road network source data by defining and calculating the aperture. Optionally, since the prediction target of the model is a small road segment of an expressway, the road network source data needs to be split or aggregated to correspond to the small road segment to obtain basic data matched with the road segment, for example, the ETC traffic data is data of each driving vehicle, and the data of the single vehicle needs to be aggregated to the corresponding prediction road segment data according to the ETC number plus the driving time; the data for lane maintenance is long-distance data, and the long-distance data needs to be divided into corresponding affected lanes of each distance. The embodiment processes the road network source data to generate the uniformly matched basic data, thereby being beneficial to processing the data in the subsequent steps and improving the efficiency.
In some of these embodiments, inputting the new features into the task regression model for prediction includes: and simultaneously predicting a plurality of traffic flow data of the specified road section of the specified time window in a multitask mode. In this embodiment, after weighting and stacking the acquired new features, the new features are input into a task regression model, a multi-task mode is adopted to perform simultaneous prediction processing on a plurality of traffic flow data such as traffic flow, average speed, traffic flow density and the like of a specified road section of a specified time window, and a traffic flow predicted value of a corresponding road section at a future time is output. Compared with the prior art that the neural network can only complete single task prediction and cannot realize simultaneous prediction of multiple road conditions, the embodiment focuses on the influence of flow caused by important road sections, emergencies and time and season factors, adopts multi-task and multi-time windows to perform simultaneous prediction, and the prediction results are beneficial to each other, such as two tasks of flow and speed, density information is obtained by calculation according to rho phi/v, the prediction results can be supplemented and influenced to achieve accuracy and optimization of the prediction results.
It should be noted that the steps illustrated in the above-described flow diagrams or in the flow diagrams of the figures may be performed in a computer system, such as a set of computer-executable instructions, and that, although a logical order is illustrated in the flow diagrams, in some cases, the steps illustrated or described may be performed in an order different than here.
The embodiment also provides a system for predicting short-term traffic flow, which is used for implementing the above embodiments and preferred embodiments, and the description of the system is omitted. As used hereinafter, the terms "module," "unit," "subunit," and the like may implement a combination of software and/or hardware for a predetermined function. Although the means described in the embodiments below are preferably implemented in software, an implementation in hardware, or a combination of software and hardware is also possible and contemplated.
Fig. 3 is a block diagram of a short-term traffic flow prediction system according to an embodiment of the present application, and as shown in fig. 3, the system includes an acquisition and preprocessing module 31, a feature extraction module 32, a feature fusion module 33, an attention mechanism module 34, and a traffic flow prediction module 35:
the acquisition and preprocessing module 31 is used for acquiring multi-modal road network source data, preprocessing the data by taking road sections and time as main dimensions, and generating basic data with uniform caliber and matched road sections; the feature extraction module 32 is configured to perform feature extraction on road network spatio-temporal data, environmental data and video image data in the basic data respectively to obtain a time-dependent feature, a space-dependent feature, an environmental feature and an image feature; a feature fusion module 33, configured to perform feature fusion on the time-dependent feature, the space-dependent feature, the environmental feature, and the image feature to obtain a fusion feature; an attention mechanism module 34, configured to filter the fused features through an attention mechanism model, and capture key valid features in the fused features; and the traffic flow prediction module 35 is configured to input the obtained effective features into the task regression model for prediction, so as to obtain a final predicted value.
By the system, the acquisition and preprocessing module 31 divides the road network data into three modes for processing aiming at the complexity and the changeability of the road network data, so that the problem of low utilization rate of the multi-mode data of the road network is solved, the utilization rate of the data is improved, and the prediction and identification of different road conditions in the follow-up process are facilitated; the feature extraction module 32 can effectively solve the problem of predicting the road condition emergency state and the multi-situation at the same time by extracting the features of the multi-mode data to obtain the time-dependent feature, the space-dependent feature, the environmental feature and the image feature, thereby being beneficial to realizing the prediction of the short-time section traffic road condition of the highway network and improving the accuracy of the subsequent prediction of different road conditions; the feature fusion module 33 performs serial fusion on the time-dependent features, the space-dependent features, the environmental features and the image features in a single-layer full-connection mode, and retains information of each type of features; the attention mechanism module 34 adds an attention mechanism model after multi-modal feature fusion, filters the fusion features according to the predicted road section targets, captures key effective features in the fusion features, and compared with the prior art that processed features are directly used as the input parameters to enter the model prediction, the embodiment predicts the key features obtained by filtering through the attention mechanism model by using the input parameters to enter the model prediction, can better reflect the real situation of high-speed short-time traffic flow, solves the problem of low accuracy of multi-situation road condition prediction, and improves the prediction accuracy of the model; the traffic flow prediction module 35 helps the high-speed management operation company to judge whether the corresponding road section in the future is smooth or not according to the predicted value of the traffic flow, and takes preventive scheduling measures for the road section which is likely to generate congestion, thereby solving the problem of predicting the sudden situation of short-distance complex road conditions and improving the accuracy and fineness of prediction. The whole system solves the problems of low utilization rate of multi-mode data of the road network and low accuracy rate of multi-situation road condition prediction under the condition of predicting the road condition in a short distance in the middle of an expressway, so that the short-time traffic flow prediction is more accurate, the outburst of the short-distance road condition is better predicted, and the accuracy and the efficiency of model prediction are improved.
In some of these embodiments, filtering the fused features through the attention mechanism model in the attention mechanism module 34 includes: establishing attention weight among the fusion features, generating an attention weight matrix, carrying out softmax normalization on the attention weight matrix to obtain a weight vector, carrying out weighted linear combination on the fusion features through the weight vector to obtain a feature sequence with a weight reference, wherein the feature sequence comprises: the method comprises the steps of adopting a space attention mechanism to capture space characteristic sequences of special road sections such as abnormal road sections, accident-prone road sections and confluence parts and adopting a time attention mechanism to capture time characteristic sequences of special time periods such as morning and evening peaks, holidays and seasonal regularity information. In the embodiment, the fusion features are filtered through the attention mechanism model to obtain key effective time dependence features, space dependence features, environmental features and image features, compared with the prior art center, the processed features are directly used as the input parameters to enter the model prediction, the embodiment can reflect the real situation of high-speed short-time traffic flow, the problem of low accuracy of road condition multi-situation prediction is solved, and the accuracy of model prediction is improved.
The present invention will be described in detail with reference to the following application scenarios.
Based on multi-mode data under multi-device multi-dimensional acquisition, the invention carries out feature fusion after extracting multi-mode data features, then adopts an attention mechanism model to obtain the features of different road sections and different prediction targets and predict the traffic flow situation of each road section in a certain period of time in the future, and fig. 4 is a data flow schematic diagram for short-time traffic flow prediction according to the embodiment of the application, as shown in fig. 4, the specific flow steps of the technical scheme for short-time traffic flow prediction in the embodiment are as follows:
s1, acquiring an ETC number, a license plate color, running time and a running direction in ETC traffic data from an ETC portal system based on service understanding and target road section prediction of source data; acquiring road section, pile number, ETC gate frame number, belonging road network information, time and weather data in road condition environment data; acquiring video shooting time, shooting position and shooting picture in the video image data;
s2, with the road section plus time as a main dimension, aggregating, splitting, converting and aligning the acquired traffic data, environment data and image data to generate basic data with uniform caliber and matched with the road section, for example, with the road section and time as the main dimension, aggregating the traffic data to generate two basic data of traffic flow and speed; converting and aligning the environmental data to generate three basic data of weather severity, whether a junction is existed or not and whether a toll station is existed or not, wherein 1 and 2 represent the weather severity; 1 represents a pivot, 0 represents not a pivot; 1 indicates a toll station, and 0 indicates not a toll station; converting and aligning the image data to generate basic data of a shot picture;
s3, extracting the road network data by utilizing the T-GCN characteristics to obtain time dependence characteristics and space dependence characteristics, and directly extracting the environmental data by adopting one-hot characteristics; performing image feature extraction on video image data by adopting a convolutional neural network, for example, taking a road segment 1 as a main dimension, and extracting time dependent features X11, X21, a space dependent feature X31 and the like from the road network data by utilizing T-GCN; extracting environmental characteristics X11, X21 and the like from environmental data by adopting one-hot; extracting image characteristics X11, X21 and the like from video image data by adopting a convolutional neural network;
s4, performing feature fusion on the extracted time-dependent features, space-dependent features, environmental features and image features, and performing combination connection on all the features in a single-layer full-connection mode, for example, performing feature fusion on the extracted time-dependent features, space-dependent features, environmental features and image features by taking a road section 1 as a main dimension to obtain fusion features W11, W21, W31 and the like;
and S5, adding an attention mechanism model into the fused features, learning historical sequence features, capturing key effective features, inputting the captured new features into a final LSTM model as a new input layer for prediction to obtain predicted values, wherein the predicted values are obtained by capturing the key effective features through the attention mechanism model by taking the road section 1 as a main dimension, inputting the key effective features into the final LSTM model for prediction to obtain predicted values y11, y21, y31 and the like.
The present embodiment also provides an electronic device comprising a memory having a computer program stored therein and a processor configured to execute the computer program to perform the steps of any of the above method embodiments.
Optionally, the electronic apparatus may further include a transmission device and an input/output device, wherein the transmission device is connected to the processor, and the input/output device is connected to the processor.
In addition, in combination with the method for predicting short-time traffic flow in the foregoing embodiments, the embodiments of the present application may provide a storage medium to implement. The storage medium having stored thereon a computer program; the computer program, when executed by a processor, implements any of the methods of short-time traffic flow prediction in the above embodiments.
In one embodiment, fig. 5 is a schematic diagram of an internal structure of an electronic device according to an embodiment of the present application, and as shown in fig. 5, an electronic device is provided, where the electronic device may be a server, and the internal structure diagram may be as shown in fig. 5. The electronic device includes a processor, a memory, a network interface, and a database connected by a system bus. Wherein the processor of the electronic device is configured to provide computing and control capabilities. The memory of the electronic equipment comprises a nonvolatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The database of the electronic device is used for storing data. The network interface of the electronic device is used for connecting and communicating with an external terminal through a network. The computer program is executed by a processor to implement a method of short-time traffic flow prediction.
Those skilled in the art will appreciate that the configuration shown in fig. 5 is a block diagram of only a portion of the configuration associated with the present application, and does not constitute a limitation on the electronic device to which the present application is applied, and a particular electronic device may include more or less components than those shown in the drawings, or may combine certain components, or have a different arrangement of components.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in the embodiments provided herein may include non-volatile and/or volatile memory, among others. Non-volatile memory can include read-only memory (ROM), Programmable ROM (PROM), Electrically Programmable ROM (EPROM), Electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus Direct RAM (RDRAM), direct bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
It should be understood by those skilled in the art that various features of the above-described embodiments can be combined in any combination, and for the sake of brevity, all possible combinations of features in the above-described embodiments are not described in detail, but rather, all combinations of features which are not inconsistent with each other should be construed as being within the scope of the present disclosure.
The above-mentioned embodiments only express several embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the concept of the present application, which falls within the scope of protection of the present application. Therefore, the protection scope of the present patent shall be subject to the appended claims.
Claims (8)
1. A method of short-term traffic flow prediction, the method comprising:
acquiring multi-modal road network source data, preprocessing the data by taking a road section and time as main dimensions, and generating basic data with uniform caliber and matched road sections;
respectively extracting the features of road network spatio-temporal data, environmental data and video image data in the basic data, wherein the road network spatio-temporal data are extracted according to a T-GCN model to obtain time dependence features and space dependence features, the environmental data are extracted by one-hot features to obtain environmental features, and the video image data are extracted by convolutional neural network features to obtain image features;
performing feature fusion on the time-dependent features, the space-dependent features, the environmental features and the image features to obtain fusion features;
filtering the fusion features through an attention mechanism model, and capturing key effective features in the fusion features;
and inputting the effective characteristics into a task regression model for prediction to obtain a predicted value.
2. The method of claim 1, wherein the filtering the fused features through an attention mechanism model comprises:
establishing attention weight among the fusion features to generate an attention weight matrix;
performing softmax normalization on the attention weight matrix to obtain a weight vector;
and carrying out weighted linear combination on the fusion features through the weight vector to obtain a feature sequence with weight reference.
3. The method of claim 1, wherein generating the aperture unification and segment matching base data comprises:
defining a calculation caliber, and converting the road network source data to obtain basic data with uniform caliber;
and splitting or aggregating the road network source data based on road sections to obtain basic data matched with the road sections.
4. The method of claim 1, wherein the inputting the significant features into a task regression model for prediction comprises:
and simultaneously predicting a plurality of traffic flow data of the specified road section of the specified time window in a multitask mode.
5. A system for short-term traffic flow prediction, the system comprising:
the acquisition and preprocessing module is used for acquiring multi-modal road network source data, preprocessing the data by taking a road section and time as main dimensions, and generating basic data with uniform caliber and matched road sections;
the characteristic extraction module is used for respectively extracting the characteristics of road network space-time data, environmental data and video image data in the basic data, wherein the road network space-time data is subjected to characteristic extraction according to a T-GCN model to obtain time dependence characteristics and space dependence characteristics, the environmental data is subjected to one-hot characteristic extraction to obtain environmental characteristics, and the video image data is subjected to convolutional neural network characteristic extraction to obtain image characteristics;
the characteristic fusion module is used for carrying out characteristic fusion on the time-dependent characteristic, the space-dependent characteristic, the environment characteristic and the image characteristic to obtain a fusion characteristic;
the attention mechanism module is used for filtering the fusion features through an attention mechanism model and capturing key effective features in the fusion features;
and the traffic flow prediction module is used for inputting the effective characteristics into the task regression model for prediction to obtain a predicted value.
6. The system of claim 5, wherein the filtering the fused features through an attention mechanism model comprises:
the attention mechanism module establishes attention weights among the fusion features to generate an attention weight matrix;
performing softmax normalization on the attention weight matrix to obtain a weight vector;
and carrying out weighted linear combination on the fusion features through the weight vector to obtain a feature sequence with weight reference.
7. An electronic device comprising a memory and a processor, wherein the memory has stored therein a computer program, and the processor is configured to execute the computer program to perform the method of short-time traffic flow prediction according to any one of claims 1 to 4.
8. A storage medium having a computer program stored thereon, wherein the computer program is arranged to perform the method of short-time traffic flow prediction according to any one of claims 1 to 4 when executed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011111809.3A CN112435462B (en) | 2020-10-16 | 2020-10-16 | Method, system, electronic device and storage medium for short-time traffic flow prediction |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011111809.3A CN112435462B (en) | 2020-10-16 | 2020-10-16 | Method, system, electronic device and storage medium for short-time traffic flow prediction |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112435462A CN112435462A (en) | 2021-03-02 |
| CN112435462B true CN112435462B (en) | 2021-12-07 |
Family
ID=74694969
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011111809.3A Active CN112435462B (en) | 2020-10-16 | 2020-10-16 | Method, system, electronic device and storage medium for short-time traffic flow prediction |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112435462B (en) |
Families Citing this family (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113052661B (en) * | 2021-04-14 | 2024-04-09 | 京东科技控股股份有限公司 | Method and device for acquiring attribute information, electronic equipment and storage medium |
| CN113096405B (en) * | 2021-06-10 | 2021-09-03 | 天津所托瑞安汽车科技有限公司 | Construction method of prediction model, and vehicle accident prediction method and device |
| CN113256985B (en) * | 2021-06-28 | 2021-09-17 | 智道网联科技(北京)有限公司 | Traffic congestion prediction method and device and electronic equipment |
| CN114283590B (en) * | 2021-09-02 | 2023-03-21 | 青岛海信网络科技股份有限公司 | Traffic flow peak prediction method and device and electronic equipment |
| CN113935555B (en) * | 2021-12-15 | 2022-03-18 | 华录易云科技有限公司 | Road network structure-based situation adaptive traffic prediction method and system |
| CN114462679A (en) * | 2022-01-04 | 2022-05-10 | 广州杰赛科技股份有限公司 | Network traffic prediction method, device, equipment and medium based on deep learning |
| CN114783183B (en) * | 2022-04-15 | 2024-05-24 | 中远海运科技股份有限公司 | Traffic situation algorithm-based monitoring method and system |
| CN115392752B (en) * | 2022-09-01 | 2023-11-28 | 亿雅捷交通系统(北京)有限公司 | Subway short-time passenger flow prediction method, system, electronic equipment and storage medium |
| CN115691144B (en) * | 2023-01-03 | 2023-03-31 | 西南交通大学 | Abnormal traffic state monitoring method, device and equipment and readable storage medium |
| CN116386336B (en) * | 2023-05-29 | 2023-08-08 | 四川国蓝中天环境科技集团有限公司 | Road network traffic flow robust calculation method and system based on bayonet license plate data |
| CN116504076A (en) * | 2023-06-19 | 2023-07-28 | 贵州宏信达高新科技有限责任公司 | Expressway traffic flow prediction method based on ETC portal data |
| CN116884215B (en) * | 2023-07-14 | 2024-05-31 | 北京百度网讯科技有限公司 | Traffic state prediction and model generation methods, devices, equipment and media |
| CN117456738B (en) * | 2023-12-26 | 2024-02-27 | 四川云控交通科技有限责任公司 | Expressway traffic volume prediction method based on ETC portal data |
| CN117475641B (en) * | 2023-12-28 | 2024-03-08 | 辽宁艾特斯智能交通技术有限公司 | Method, device, equipment and medium for detecting traffic state of expressway |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109670277A (en) * | 2019-02-19 | 2019-04-23 | 南京邮电大学 | A kind of travel time prediction method based on multi-modal data fusion and multiple-model integration |
| CN110782663A (en) * | 2019-09-30 | 2020-02-11 | 电子科技大学 | Road network traffic flow short-time prediction method combining time-space characteristics |
| CN110827543A (en) * | 2019-11-11 | 2020-02-21 | 重庆邮电大学 | Short-term traffic flow control method based on deep learning and spatio-temporal data fusion |
| CN111275971A (en) * | 2020-02-18 | 2020-06-12 | 山西交通控股集团有限公司 | Expressway traffic flow prediction method |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20150072471A (en) * | 2013-12-19 | 2015-06-30 | 건국대학교 산학협력단 | Traffic flow prediction system using spatiotemporal stochastic model |
-
2020
- 2020-10-16 CN CN202011111809.3A patent/CN112435462B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109670277A (en) * | 2019-02-19 | 2019-04-23 | 南京邮电大学 | A kind of travel time prediction method based on multi-modal data fusion and multiple-model integration |
| CN110782663A (en) * | 2019-09-30 | 2020-02-11 | 电子科技大学 | Road network traffic flow short-time prediction method combining time-space characteristics |
| CN110827543A (en) * | 2019-11-11 | 2020-02-21 | 重庆邮电大学 | Short-term traffic flow control method based on deep learning and spatio-temporal data fusion |
| CN111275971A (en) * | 2020-02-18 | 2020-06-12 | 山西交通控股集团有限公司 | Expressway traffic flow prediction method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112435462A (en) | 2021-03-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112435462B (en) | Method, system, electronic device and storage medium for short-time traffic flow prediction | |
| CN110414451B (en) | License plate recognition method, device, equipment and storage medium based on end-to-end | |
| CN110929934A (en) | Equipment failure prediction method and device, computer equipment and storage medium | |
| Song et al. | A match‐then‐predict method for daily traffic flow forecasting based on group method of data handling | |
| Doshi et al. | Federated learning-based driver activity recognition for edge devices | |
| Xia et al. | A parallel grid-search-based SVM optimization algorithm on Spark for passenger hotspot prediction | |
| CN115131634A (en) | Image recognition method, device, equipment, storage medium and computer program product | |
| Rahman et al. | Attention based deep hybrid networks for traffic flow prediction using google maps data | |
| CN214338041U (en) | Intelligent city monitoring system based on 5G Internet of things | |
| Shinkuma et al. | Data assessment and prioritization in mobile networks for real-time prediction of spatial information using machine learning | |
| KR20240011847A (en) | Method for predicting time-variable data for weg page, apparatus, web management system using thereof, computer-readable storage medium and computer program | |
| CN112766176A (en) | Training method of lightweight convolutional neural network and face attribute recognition method | |
| CN115567563B (en) | Comprehensive transportation hub monitoring and early warning system based on end edge cloud and control method thereof | |
| CN117133124A (en) | Digital twin intelligent highway charging and management system and method | |
| CN115906927A (en) | Data access analysis method and system based on artificial intelligence and cloud platform | |
| CN115277789A (en) | Safety protection system and method for cascade hydropower station | |
| KR20230085299A (en) | System and method for detecting damage of structure by generating multi-scale resolution image | |
| Iordache et al. | Industrial Expert System for Intelligent Traffic Lane Allocation Using Machine Learning and Pattern Recognition | |
| Guo et al. | Driving Risk Assessment Based on High-frequency, High-resolution Telematics Data | |
| CN113470012B (en) | Marking identification method and device, storage medium and electronic device | |
| CN113792941B (en) | Method, device, computer equipment and storage medium for predicting road traffic speed | |
| Kannojia et al. | Meta fusion of trained heterogeneous image classifier using classifier selection | |
| CN116777621B (en) | Trade business risk prevention and control method and prevention and control system thereof | |
| CN116824277B (en) | Visual target detection model for road disease detection, construction method and application | |
| Africa | International Journal of Emerging Trends in Engineering Research |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |