CN116504076A

CN116504076A - Expressway traffic flow prediction method based on ETC portal data

Info

Publication number: CN116504076A
Application number: CN202310724716.5A
Authority: CN
Inventors: 阮雪飞; 李大全; 朱秋实; 王志海; 杨唐; 吴政沅
Original assignee: Guizhou Hongxinda High New Technology Co ltd
Current assignee: Guizhou Hongxinda High New Technology Co ltd
Priority date: 2023-06-19
Filing date: 2023-06-19
Publication date: 2023-07-28

Abstract

The invention discloses an ETC portal data-based expressway traffic flow prediction method, which comprises the following steps: acquiring ETC portal data of the expressway through an ETC system, preprocessing the ETC portal data by utilizing a one-dimensional convolution filter, and extracting key data of the ETC portal; inputting the extracted key data into a traffic flow early warning model based on a multidimensional deep recurrent neural network, and carrying out real-time predictive analysis on the key data of the ETC portal to obtain a traffic flow prediction result; and (3) formulating a traffic control scheme based on the traffic flow prediction result, and performing management measures on traffic according to the traffic control scheme, wherein the management measures comprise intelligent traffic scheduling, automatic command and congestion control measures. The traffic flow prediction accuracy can be improved, the road traffic efficiency is improved, an efficient and accurate management means is provided for traffic management departments, and convenience is provided for road users.

Description

Expressway traffic flow prediction method based on ETC portal data

Technical Field

The invention relates to the technical field of traffic flow prediction, in particular to an expressway traffic flow prediction method based on ETC portal data.

Background

In recent years, the operation mileage of the expressway is increased year by year, the road network density is continuously improved, the traffic flow is continuously increased, and the corresponding problems of operation efficiency, traffic safety and operation management are needed to be solved. Meanwhile, as the standard construction of the traffic transportation road network monitoring system in China is started later, the depth and breadth of the research on the standard construction of the road network monitoring system are insufficient, the autonomy and innovation of the related theoretical research are insufficient, the relevant regulation system of the road network monitoring management is absent, the level of the legislation is not high, the national standard and industry standard construction related to the road network monitoring management of the trunk road network are not perfect, the requirement of the actual road network monitoring work is difficult to support, and the problems that the theoretical research is lagged behind the current development state, the standard basis of the road network operation evaluation is insufficient, the standardization degree of the road network management work is not high and the like exist.

Currently, predicting the traffic flow of a highway in real time becomes one of important tasks of traffic management, and accurate traffic flow prediction can provide effective support for traffic monitoring, congestion control and road safety. However, most existing traffic flow prediction methods are based on fixed sensor data, such as traffic monitoring cameras and coil detectors, and have large data acquisition limitations and low accuracy.

Disclosure of Invention

The invention provides an ETC portal data-based expressway traffic flow prediction method, which aims to solve the problem that the real-time expressway traffic flow prediction in the prior art becomes one of important tasks of traffic management, and the accurate traffic flow prediction can provide effective support for traffic monitoring, congestion control and road safety. However, most of the existing traffic flow prediction methods are based on fixed sensor data, such as traffic monitoring cameras and coil detectors, and have the problems of large data acquisition limitation and low accuracy.

In order to achieve the above purpose, the present invention provides the following technical solutions:

the highway traffic flow prediction method based on ETC portal data comprises the following steps:

s101: acquiring ETC portal data of the expressway through an ETC system, preprocessing the ETC portal data by utilizing a one-dimensional convolution filter, and extracting key data of the ETC portal;

s102: inputting the extracted key data into a traffic flow early warning model based on a multidimensional deep recurrent neural network, and carrying out real-time predictive analysis on the key data of the ETC portal to obtain a traffic flow prediction result;

s103: and (3) formulating a traffic control scheme based on the traffic flow prediction result, and performing management measures on traffic according to the traffic control scheme, wherein the management measures comprise intelligent traffic scheduling, automatic command and congestion control measures.

Wherein, the step S101 includes:

s1011: setting an ETC portal frame on a specific road section of a highway, connecting with an ETC system, receiving signals passed by vehicles, converting the received signals passed by the vehicles into ETC portal frame data, and transmitting the ETC portal frame data acquired in real time to a data processing center for data processing through a wireless sensor network;

s1012: preprocessing ETC portal data in a data processing center through a self-adaptive one-dimensional discrete convolution filtering algorithm, filtering noise signals in the ETC portal data based on a mode that dynamic noise estimation replaces a fixed convolution kernel, and obtaining preprocessed ETC portal data;

s1013: and automatically identifying and extracting key fields from the preprocessed ETC portal data by using an association rule mining algorithm, and acquiring ETC portal key data, wherein the ETC portal key data comprises vehicle passing time, vehicle identification numbers, vehicle types, vehicle speeds and lane occupancy rate.

Wherein, the step S102 includes:

s1021: based on an ETC portal data management library, a plurality of pieces of running water ETC portal data are called, the running water ETC portal data are divided into a training set and a testing set by utilizing a self-adaptive data division strategy, wherein the training set is used for training a multidimensional deep recurrent neural network model, and the testing set is used for evaluating the performance of the model;

S1022: training the multidimensional deep recurrent neural network by using the training set by taking the space-time characteristics and the external factor characteristics in the training set as inputs, and learning the change rule of the traffic flow in multiple dimensions by the multidimensional deep recurrent neural network; using an attention mechanism to adaptively allocate weights, acquiring the attention degree of the multidimensional deep recurrent neural network to key data, and constructing a traffic flow early warning model based on the trained multidimensional deep recurrent neural network;

s1023: inputting current ETC portal key data into a traffic flow early warning model, outputting a traffic flow prediction result, wherein the traffic flow prediction result comprises traffic flow early warning signals and key factors causing flow mutation, and providing targeted decision advice for traffic management departments by analyzing the early warning signals and the key factors in real time; the traffic flow early warning signal is set based on real-time traffic information of traffic flow, speed and lane occupation, and comprises normal flow, congestion flow, serious congestion and extreme congestion, and key factors causing abrupt flow change comprise traffic accidents, road construction and bad weather.

Wherein, the step S103 includes:

S1031: according to the traffic flow prediction result and the actual traffic condition, adopting an intelligent optimization algorithm to formulate a traffic control scheme, wherein the traffic control scheme comprises traffic scheduling, automatic command and congestion control measures which are automatically adjusted according to the prediction result;

s1032: according to the traffic control scheme, corresponding limiting measures are adopted for different vehicles and different lanes, when congestion occurs in a road section, the control measures of a high-speed entrance and an emergency lane are automatically controlled and started through a real-time monitoring and rear-end control system coordinated across areas, the congestion is eliminated, and meanwhile dynamic information display of a high-speed road section is started to guide a driver to reasonably plan a journey;

s1033: the effect of implementing management measures and traffic control schemes is monitored and evaluated in real time through big data and machine learning technology, the control strategy is automatically adjusted according to the evaluation result, intelligent prediction is carried out on future flow data, and continuous optimization of a traffic flow early warning model is realized; the big data technology refers to the storage and processing of real-time traffic data by using a distributed computing framework, and the machine learning technology refers to the modeling of traffic flow prediction by using a time sequence analysis model.

Wherein, the step S1011 includes:

The method comprises the steps of converting received signals of vehicle passing into ETC portal data, and acquiring the signals of vehicle passing, wherein the signals of vehicle passing comprise signals sent by a vehicle-mounted device carried by a vehicle and signals sent by equipment for acquiring vehicle running information;

decoding a signal which a vehicle passes through, extracting corresponding vehicle identification information through an intelligent vehicle identification information recognition algorithm model, wherein the vehicle identification information comprises license plate numbers and vehicle type information, and screening, verifying and encrypting the vehicle identification information, wherein the intelligent vehicle identification information recognition algorithm is used for recognizing and extracting the license plate numbers and the vehicle type information by utilizing an image recognition algorithm of a two-dimensional convolutional neural network;

inputting the vehicle identification information into an ETC portal data generation algorithm model, acquiring ETC portal data comprising the vehicle identification information, and storing the ETC portal data into an ETC portal data management library;

and continuously optimizing the vehicle identification information recognition algorithm model, the ETC portal data generation algorithm model and the corresponding rules through a machine learning algorithm and a big data analysis technology.

Wherein, step S1013 includes:

in the process of automatically identifying and extracting key fields from the preprocessed ETC portal data by using an association rule mining algorithm, converting the preprocessed ETC portal data into a format required by the association rule mining algorithm, wherein the format comprises a standardized format and a binary coding format;

The method comprises the steps of carrying out frequent item set mining on an ETC portal data set to obtain all frequently occurring item sets, wherein the frequent item sets are combinations of items with the occurrence frequency higher than a given threshold value;

generating association rules based on the frequent item sets; calculating indexes of the support degree, the confidence degree and the lifting degree of each association rule, and evaluating the quality of the association rule;

and acquiring key fields according to indexes of the support degree, the confidence degree and the lifting degree of the association rule, wherein the key fields refer to attributes with important characteristics in ETC portal data.

Wherein, step S1022 includes:

in the process of constructing a traffic flow early warning model based on a trained multidimensional depth recurrent neural network, preprocessing ETC portal critical data acquired in real time, including data cleaning and data normalization processing, inputting the preprocessed ETC portal critical data into the trained multidimensional depth recurrent neural network, predicting future traffic flow through the trained multidimensional depth recurrent neural network, according to a traffic flow prediction result, the prediction result comprises ETC portal critical data, analyzing historical data to acquire traffic management requirements of a current period, setting a threshold value of the predicted traffic flow according to the traffic management requirements of the current period, comparing the ETC portal critical data with the threshold value, judging whether the current traffic condition reaches an early warning condition, wherein the threshold value of the predicted traffic flow is set, and judging that the early warning condition is reached when the ETC portal critical data exceeds the set threshold value; if the early warning condition is met, entering a traffic control stage, otherwise, continuing monitoring; when the early warning condition is met, the traffic flow early warning model automatically sends early warning information to related departments, the related departments are reminded of keeping vigilance to traffic jam conditions, and corresponding traffic management measures are made.

Wherein, the step S1031 includes:

in the process of adopting an intelligent optimization algorithm to formulate a traffic control scheme according to a traffic flow prediction result and a real traffic condition, a traffic control scheme is represented by chromosome codes, and the traffic control scheme comprises a specific control period, control measures and control points; initializing a population, and randomly generating a preset number of individuals, wherein the individuals are traffic control schemes; calculating individual fitness, and evaluating according to individual control measures and objective functions, wherein the objective functions comprise the degree of congestion relief and traffic flow capacity; selecting individuals in the population, and selecting a preset number of individuals as parents; performing genetic manipulation on the parent, including crossover and mutation, to produce offspring individuals; calculating the fitness of the offspring individuals, eliminating, and reserving the individuals with the best fitness to replace some individuals in the population; judging whether the condition of stopping iteration is met, and if the preset iteration times or the fitness meets a certain requirement, outputting a traffic control scheme with the best final fitness.

Wherein, in step S1032, the real-time monitoring and back-end control system for cross-region coordination automatically controls the high-speed access and opening the control measures of the emergency lane, including:

Corresponding monitoring equipment including cameras and sensors are arranged at the entrance and the exit of the expressway and at the key positions of the expressway to monitor the conditions of vehicles and traffic, and meanwhile, a rear-end control system is arranged to receive and process data acquired by the monitoring equipment; setting up a highway toll station and a traffic command center, transmitting acquired data to a rear-end control system, carrying out data processing and analysis, predicting traffic conditions of road sections and intersections, and automatically setting corresponding control instructions by the rear-end control system based on the predicted traffic conditions, wherein the control instructions comprise starting an emergency lane of a specific area and limiting the speed of the specific road section, and the control instructions are issued in a designated mode, wherein the designated mode comprises a signpost, an LED drive-by-wire screen, an electronic police and a display screen; when the traffic condition changes, the back-end control system automatically adjusts control instructions in real time, including opening other emergency lanes; the expressway management center acquires traffic control condition feedback in real time through the mobile phone app, and modifies and perfects a traffic control scheme through feedback information.

The method for mining the frequent item sets of the ETC portal data sets comprises the following steps of:

Setting a threshold value of frequent item set mining according to the domain knowledge; generating all corresponding item sets from the ETC portal data set, and obtaining candidate item sets through combination and pruning; scanning the data set, counting the occurrence times of each candidate item set, and generating a frequent item set; generating an association rule based on the frequent item set, wherein the association rule comprises a front piece and a back piece, and the front piece and the back piece are item sets;

the process of generating the association rule includes: setting a confidence threshold generated by the association rule according to the actual situation; for each frequent item set, generating an association rule that neither the front part nor the back part is empty, and calculating the confidence coefficient of the association rule; deleting rules with confidence coefficient not reaching a threshold value; for rules for which the front piece is not empty, new association rules are recursively mined.

Compared with the prior art, the invention has the following advantages:

the highway traffic flow prediction method based on ETC portal data comprises the following steps: acquiring ETC portal data of the expressway through an ETC system, preprocessing the ETC portal data by utilizing a one-dimensional convolution filter, and extracting key data of the ETC portal; inputting the extracted key data into a traffic flow early warning model based on a multidimensional deep recurrent neural network, and carrying out real-time predictive analysis on the key data of the ETC portal to obtain a traffic flow prediction result; and (3) formulating a traffic control scheme based on the traffic flow prediction result, and performing management measures on traffic according to the traffic control scheme, wherein the management measures comprise intelligent traffic scheduling, automatic command and congestion control measures. The traffic flow prediction accuracy can be improved, the road traffic efficiency is improved, an efficient and accurate management means is provided for traffic management departments, and convenience is provided for road users.

Additional features and advantages of the invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention.

The technical scheme of the invention is further described in detail through the drawings and the embodiments.

Drawings

The accompanying drawings are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate the invention and together with the embodiments of the invention, serve to explain the invention. In the drawings:

FIG. 1 is a flow chart of a highway traffic prediction method based on ETC portal data in an embodiment of the invention;

FIG. 2 is a flow chart of extracting key data of an ETC portal in an embodiment of the invention;

FIG. 3 is a flow chart of obtaining a traffic flow prediction result in an embodiment of the present invention.

Detailed Description

The preferred embodiments of the present invention will be described below with reference to the accompanying drawings, it being understood that the preferred embodiments described herein are for illustration and explanation of the present invention only, and are not intended to limit the present invention.

The embodiment of the invention provides an expressway traffic flow prediction method based on ETC portal data, which comprises the following steps:

The working principle of the technical scheme is as follows: the method is based on massive running water data generated by a highway networking charging system, a one-dimensional convolution filter is applied to input data feature extraction through predictive analysis of a neural network algorithm, a traffic flow early warning model is constructed, estimation of future road traffic volume is provided for traffic management departments and road users, and theoretical basis is provided for road management and guidance of traffic.

Firstly, acquiring ETC portal data of an expressway in real time through an ETC system, wherein the data comprise vehicle entrance time, exit time, place, vehicle Identification Number (VIN), license plate, vehicle type, lane information, vehicle speed, lane occupancy rate, charging grade and the like; after preprocessing operation, extracting ETC portal key data, wherein the key data comprise traffic flow, vehicle distribution, vehicle type distribution and the like, and the data are helpful for traffic flow prediction and traffic control;

Then, the extracted key data are input into a traffic flow early warning model based on a multidimensional deep recurrent neural network, the multidimensional deep recurrent neural network has strong feature extraction capability, long-term dependency in the data can be captured, complex modes and features in the data can be effectively learned, and therefore real-time prediction analysis of high-precision ETC portal key data is realized; obtaining a traffic flow prediction result through a deep recurrent neural network model, wherein the prediction result comprises information such as traffic flow trend, congestion area and the like in a short period, which is helpful for a traffic management center to make corresponding measures to deal with congestion conditions in advance;

finally, based on the traffic flow prediction result, a reasonable traffic control scheme is formulated; corresponding intelligent traffic scheduling, automatic command and congestion control measures are adopted according to a traffic control scheme, so that real-time traffic monitoring and congestion relief of the expressway are realized;

the method can realize real-time monitoring of the traffic flow of the expressway and timely find out potential congestion areas, so that targeted traffic control measures are adopted, congestion duration is reduced, road traffic capacity is improved, and finally intelligent management of expressway traffic is realized.

The beneficial effects of the technical scheme are as follows: accurately predicting the road traffic flow of a specific time period in the future by utilizing a one-dimensional convolution filter and a multidimensional deep recurrent neural network; the traffic control scheme formulated based on the traffic flow prediction result has individuation and dynamic adaptability, and can better meet the requirements of different roads, time periods and traffic conditions; by adopting intelligent traffic scheduling, automatic command and advanced congestion control technology, more efficient traffic management and road traffic efficiency can be realized; the model supports real-time predictive analysis, can quickly respond to traffic condition changes, and provides timely and effective decision basis for traffic management departments.

In another embodiment, the step S101 includes:

The working principle of the technical scheme is as follows: firstly, setting an ETC portal frame on a specific road section of a highway, connecting the ETC portal frame with an ETC system, receiving signals passed by vehicles, converting the received signals passed by the vehicles into ETC portal frame data, and transmitting the ETC portal frame data acquired in real time to a data processing center for data processing through a wireless sensor network; then, preprocessing ETC portal data by a one-dimensional discrete convolution filtering algorithm, filtering noise signals in the ETC portal data based on a mode that dynamic noise estimation replaces a fixed convolution kernel, and acquiring preprocessed ETC portal data, wherein the aim of the step is to improve data quality and enable subsequent data analysis to be more accurate; finally, key fields are automatically identified and extracted from the preprocessed ETC portal data by utilizing an association rule mining algorithm, so that ETC portal key data are obtained, wherein the ETC portal key data comprise vehicle passing time, vehicle identification numbers, vehicle types, vehicle speeds and lane occupancy rates, and the key data are helpful for vehicle flow analysis, vehicle type analysis, vehicle speed analysis, lane occupancy rate analysis and the like, and can help expressway operation management, traffic planning, safety risk control and the like to make more accurate decisions.

The pretreatment of ETC portal data comprises the following steps: performing de-duplication operation on the vehicle identification, the portal identification and the time, and avoiding the condition that the ETC portal data have vehicles passing through the same portal at different times; carrying out reasonable rule filling on the missing portal frame identification and time information; and converting the time into a standard format, so that the subsequent processing is convenient.

The beneficial effects of the technical scheme are as follows: the ETC portal data are acquired in real time, so that timeliness and usability of the data are improved; preprocessing is carried out through a discrete convolution filtering algorithm, so that the data quality is improved, and the subsequent data analysis is more accurate; the key fields are automatically identified and extracted by adopting an automatic association rule mining algorithm, so that the ETC portal key data are obtained, and the method is favorable for carrying out traffic flow analysis, vehicle type analysis, vehicle speed analysis, lane occupancy analysis and the like, and provides scientific basis for expressway operation management, traffic planning, safety risk control and the like.

In another embodiment, the step S102 includes:

The working principle of the technical scheme is as follows: the training set and the testing set are divided by utilizing the self-adaptive data division strategy, so that the model can fully adjust parameters to adapt to various actual scenes; extracting space-time characteristics and external factor characteristics in a training set as inputs, so that a model can grasp key modes in data; an attention mechanism is introduced in the model training process, so that the model can adaptively distribute weights, and key influencing factors are fully focused; and outputting a prediction result comprising traffic flow early warning signals and key factors possibly causing flow mutation, and helping traffic management departments to make scientific decisions. Wherein, traffic flow early warning signal: in general, the early warning signals can be classified into a plurality of levels, including green (normal flow), yellow (crowded flow), orange (severe crowded), red (extremely crowded) and the like, and the setting of the early warning signals is based on real-time traffic information such as traffic flow, speed, lane occupation and the like; key factors leading to abrupt flow changes: including traffic accidents, road construction, bad weather (rain, snow, fog, etc.).

The beneficial effects of the technical scheme are as follows: the partition rationality of the training set and the test set is improved, the data samples are fully utilized, and the generalization capability of the model is improved; by introducing space-time characteristics and external factor characteristics, the model can better capture the change rule of the actual traffic flow; adopting an attention mechanism, improving the attention of the model to key influence factors, and optimizing the model learning process; the output early warning signals and key influencing factors provide valuable information for traffic management departments and road users, are beneficial to improving the accuracy of traffic condition prediction, and realize rapid and effective traffic condition improvement.

In another embodiment, the step S103 includes:

S1033: real-time monitoring and evaluation are carried out on the effect of implementing management measures and traffic control schemes by using big data and machine learning technology, a control strategy is automatically adjusted according to an evaluation result, intelligent prediction is carried out on future flow data, and continuous optimization of a traffic flow early warning model is realized; the big data technology refers to the storage and processing of real-time traffic data by using a distributed computing framework, and the machine learning technology refers to the modeling of traffic flow prediction by using a time sequence analysis model.

The working principle of the technical scheme is as follows: based on the traffic flow prediction result, an intelligent optimization algorithm is adopted to formulate a traffic control scheme to realize the monitoring and control of traffic conditions, and meanwhile, big data and machine learning technology are combined to monitor and evaluate the effect of the traffic control scheme in real time so as to intelligently predict future flow data. Firstly, adopting an intelligent optimization algorithm to formulate a traffic control scheme through a traffic flow prediction result and a real traffic condition, wherein the traffic control scheme comprises various management measures such as traffic scheduling, automatic command and congestion control measures; secondly, according to traffic control schemes, corresponding limiting measures are adopted for different vehicles and different lanes, when congestion occurs in road sections, the control measures of a high-speed entrance and an emergency lane are automatically controlled through a cross-regional coordinated real-time monitoring and rear-end control system, the congestion is eliminated, dynamic information display of a high-speed road section is started, a driver is guided to reasonably plan a journey, and the measures aim to improve the traffic efficiency and safety of the expressway. Finally, by applying big data and machine learning technology, the effect of implementing management measures and traffic control schemes is monitored and evaluated in real time, and the control strategy is automatically adjusted and the future flow data is intelligently predicted according to the evaluation result. The time sequence analysis model is an algorithm in a machine learning technology, a numerical value of a certain time point in the future is predicted by modeling time sequence data, the time sequence analysis model is built, historical traffic flow data is required to be acquired, the historical traffic flow data is processed and analyzed, input and output of a relevant time sequence model are determined, a corresponding prediction model is built, after the latest traffic data is received, the traffic flow in a certain time period in the future is predicted through the prediction model, and corresponding management decisions are made.

The beneficial effects of the technical scheme are as follows: the traffic condition is monitored and dealt with in time, the refinement and intelligence level of traffic management is improved, the risks of traffic jam and accidents are reduced, and meanwhile the traffic capacity of roads and the safety of a traffic system are improved.

In another embodiment, the step S1011 includes:

the method comprises the steps of converting received signals of vehicle passing into ETC portal data, and acquiring the signals of vehicle passing, wherein the signals of vehicle passing comprise signals sent by a vehicle-mounted device carried by a vehicle or signals sent by equipment for acquiring vehicle running information;

The working principle of the technical scheme is as follows: the ETC portal frame needs to acquire vehicle passing signals to determine the actual traffic situation and charging information of the vehicle, and the vehicle sends signals by carrying vehicle-mounted devices or acquires vehicle running information through portal frame equipment to send signals; the portal frame is used for decoding the vehicle passing signals, extracting vehicle identification information, including license plate numbers and vehicle type information, automatically identifying and extracting the vehicle identification information by adopting an intelligent vehicle identification information identification algorithm model, so that the accuracy and the efficiency can be greatly improved, and meanwhile, the operation and maintenance cost of the portal frame is reduced; the portal frame needs to screen, check and encrypt the vehicle identification information, so that the safety and reliability of the passing vehicle information are ensured, information leakage and data tampering are avoided, the encryption algorithm is adopted to encrypt the vehicle identification information, the risks of information leakage and data tampering are reduced, and the data security is improved; the portal needs to generate ETC portal data comprising vehicle identification information based on the vehicle identification information, and store the ETC portal data into an ETC portal data management database, wherein the ETC portal data comprises information such as vehicle passing time, portal position, travel distance, toll amount and the like, so that follow-up financial accounting and vehicle management are facilitated; in order to improve the intelligent, humanized and benefit levels of portal equipment, an algorithm model and rules for identifying vehicle identification information and generating ETC portal data are required to be continuously iterated and optimized through a machine learning algorithm and a big data analysis technology so as to improve the accuracy and efficiency of the system and reduce the maintenance cost;

In the process of screening, verifying and encrypting the vehicle identification information, firstly, the vehicle identification information is screened through a series of rules to determine the information needing to be verified and encrypted, and the series of rules comprise screening the vehicle according to the information of the type, license plate number, vehicle color and registration place of the vehicle. Checking the selected vehicle identification information to ensure the accuracy and legality of the vehicle identification information, comparing and verifying the vehicle identification information with the data in the vehicle information database disclosed by the related departments to ensure that false conditions do not exist in the vehicle identification information, if inconsistent places are found, processing in time, and allowing the checked information to be subjected to subsequent encryption processing; the vehicle identification information after verification is encrypted, and a pair of public keys and private keys are used for encryption and decryption. Finally, marking the encrypted vehicle identification information so as to facilitate subsequent tracking and verification; for example, the encrypted information is preceded by a specific tag that enables it to be identified and queried in the database.

The beneficial effects of the technical scheme are as follows: by adopting artificial intelligence technologies such as deep learning, convolutional neural network, natural language processing and the like, the intelligent efficient processing of the vehicle identification information is realized, the recognition rate and accuracy can be greatly improved, and meanwhile, the operation efficiency and profit of portal equipment are improved; the vehicle identification information is encrypted by adopting an encryption algorithm, so that the data security is improved, and the risks of information leakage and data tampering are reduced; through a machine learning algorithm and a data analysis technology, iterative optimization of an algorithm model and rules for identifying vehicle identification information and generating ETC portal data is carried out, so that the accuracy and efficiency of a system are improved, the maintenance cost is reduced, and the intellectualization, humanization and benefit of portal equipment are realized; by generating ETC portal data, real-time recording and statistics of vehicle traffic conditions are realized, the efficiency and accuracy of vehicle management and financial accounting can be greatly improved, and the operation cost is reduced.

In another embodiment, step S1013 includes:

The working principle of the technical scheme is as follows: converting the preprocessed ETC portal data into a format required by an association rule mining algorithm, wherein the format comprises a standardized format (all portal identifications passed by each vehicle are formed into a sequence according to time sequence and used as standardized data formats) and a binary coding format (the portal identifications are converted into binary codes to indicate whether the vehicles pass by the portal or not, and each vehicle generates a binary coding sequence as data in the binary coding format);

Frequent item set mining is performed on the data set after format conversion, and all frequently occurring item sets are obtained, wherein the frequent item sets are combinations of items with the occurrence frequency higher than a given threshold value, and the method comprises the following steps:

(1) Setting a threshold value: setting a threshold value of frequent item set mining according to experience or domain knowledge;

(2) Generating a candidate item set: generating all possible item sets from the data set, and obtaining candidate item sets through combination and pruning;

(3) Counting the occurrence times: scanning the data sets, counting the occurrence times of each candidate item set, and generating frequent item sets.

Based on the frequent item set, an association rule can be generated, wherein the association rule comprises a front piece and a back piece, the front piece and the back piece are both item sets, and the step of generating the association rule is as follows:

(1) Setting a confidence threshold: setting a confidence threshold generated by the association rule according to the actual situation;

(2) Generating a strong association rule: for each frequent item set, generating an association rule that neither the front part nor the back part is empty, and calculating the confidence coefficient of the association rule;

(3) Pruning rules: deleting rules with confidence coefficient not reaching a threshold value;

(4) And (3) recursively excavating: for rules for which the front piece is not empty, new association rules are recursively mined.

Calculating the support, confidence and promotion index of each generated association rule for evaluating the quality of the association rule, wherein the support refers to the intersection probability of all item sets in the association rule; confidence refers to the probability of containing the back part B at the same time in the transaction containing the front part A; the lifting degree refers to the lifting times of the occurrence probability of the back part B in the transaction comprising the front part A compared with the occurrence probability of the back part B in the transaction without the front part A;

according to the supportability, confidence and promotion index of the association rule, a key field is obtained, wherein the key field refers to an attribute with important characteristics in ETC portal data, screening is carried out according to different indexes, the attribute with high score is selected as the key field, and for the association rule with higher promotion, the front piece attribute can be used as the key field.

The beneficial effects of the technical scheme are as follows: according to the scheme, key fields can be automatically extracted from ETC portal data, and important features in the data can be rapidly positioned, so that the data can be rapidly and accurately analyzed and processed, and the efficiency and accuracy of data analysis are improved; meanwhile, the scheme is suitable for ETC portal data sets with different scales and formats, and has strong universality and flexibility.

In another embodiment, the step S1022 includes:

The working principle of the technical scheme is as follows: acquiring ETC portal key data in real time, wherein the key data comprise traffic flow, vehicle speed, lane occupancy, vehicle type distribution and the like, and the key data are used as input data for traffic flow prediction; preprocessing ETC portal key data collected in real time, including data cleaning (deleting abnormal values, denoising and the like) and data normalization (eliminating dimension influence, so that data with different attributes have comparability); inputting the preprocessed ETC portal key data into a trained Multidimensional Deep Recurrent Neural Network (MDRNN), and predicting future traffic flow by utilizing the change rule of the traffic flow learned by the model in a plurality of dimensions; judging whether the current traffic condition reaches the early warning condition according to the prediction result by combining the historical data and the traffic management requirement, setting a threshold value of the predicted traffic flow, for example, judging that the traffic flow exceeds a certain set value or the traffic congestion index exceeds a certain set range, and judging that the early warning condition is reached; if the early warning condition is met, entering a traffic control stage; if not, continuing to monitor; when the early warning condition is met, the traffic flow early warning model automatically sends early warning information to related departments, reminds the departments of keeping vigilance to traffic jam conditions, and adopts corresponding traffic management measures, such as opening an emergency lane or issuing traffic jam information to drivers.

Where traffic management requirements refer to different time periods or are faced with key factors that lead to abrupt changes in flow, the requirements for current ETC portal key data are different.

For example: in the early peak period of a certain city, the critical data of a certain ETC portal frame shows that the traffic flow rises sharply and exceeds a set threshold value. Through predictive analysis of the traffic flow early warning model, the possibility of traffic jam of the road section within 30 minutes in the future is high. At this time, the early warning model can automatically send early warning information to the traffic management department, and take targeted management measures, such as opening an emergency lane in advance, so as to relieve the congestion condition.

The beneficial effects of the technical scheme are as follows: early warning of traffic jam conditions is performed in advance, and early warning service is provided for traffic management departments and citizens; the key data of the ETC portal are collected in real time, so that the traffic jam situation is quickly reflected, traffic management measures are timely made, and the smooth urban traffic is promoted; and the traffic flow prediction is performed by using the deep recurrent neural network, so that the accuracy of the prediction is improved, and more scientific decision support is provided for traffic management.

In another embodiment, the step S1031 includes:

The working principle of the technical scheme is as follows: the optimization of the traffic control scheme is converted into an optimization problem through intelligent optimization algorithms such as a genetic algorithm, and a better control scheme is generated through continuous optimization so as to achieve the aim; the traffic management scheme is formulated by using the intelligent optimization algorithm, so that the road traffic efficiency can be greatly improved, traffic jam is relieved, and the utilization of public resources is better realized.

When congestion occurs on the expressway, the intelligent optimization algorithm can rapidly simulate the traffic flow change trend and the road traffic condition, and traffic management measures are dynamically adjusted according to real-time traffic information to relieve the congestion as much as possible. And (3) obtaining a batch of potential schemes through screening of an optimization algorithm, and continuously generating more excellent solutions through crossover and mutation operations until the optimal scheme is found. Finally, the highway traffic management department successfully formulates a traffic control scheme, so that the highway traffic jam problem is effectively relieved while the traffic running efficiency is ensured, and the highway traffic running efficiency is improved.

The beneficial effects of the technical scheme are as follows: by adopting an intelligent optimization algorithm to formulate a traffic control scheme, the traffic control scheme can be rapidly optimized according to the traffic flow prediction result and the actual traffic condition, so that the optimal traffic efficiency is achieved, and the traffic jam condition is effectively relieved.

In another embodiment, the step S1032 includes:

in the process of controlling the entrance and exit of the expressway and opening the emergency lane automatically through the real-time monitoring and rear-end control system which is coordinated across areas, corresponding monitoring equipment is arranged on the entrance and exit of the expressway and the high-speed key positions, the corresponding monitoring equipment comprises cameras and sensors, the vehicle and traffic conditions are monitored, and meanwhile, the rear-end control system is arranged to receive and process data acquired by the monitoring equipment; setting up a highway toll station and a traffic command center, transmitting acquired data to a rear-end control system, carrying out data processing and analysis, predicting traffic conditions of road sections and intersections, and automatically setting corresponding control instructions by the rear-end control system based on the predicted traffic conditions, wherein the control instructions comprise starting an emergency lane of a specific area and limiting the speed of the specific road section, and the control instructions are issued in a designated mode, wherein the designated mode comprises a signpost, an LED drive-by-wire screen, an electronic police and a display screen; when the traffic condition changes, the back-end control system automatically adjusts control instructions in real time, including opening other emergency lanes; the expressway management center acquires traffic control condition feedback in real time through the mobile phone app, and modifies and perfects a traffic control scheme through feedback information.

The working principle of the technical scheme is as follows: monitoring equipment (such as cameras and sensors) at the entrance, the exit and the key positions of the expressway collect the vehicle and the traffic condition in real time and transmit real-time data to a rear-end control system; the back-end control system processes and analyzes the data transmitted in real time, predicts future traffic conditions of road sections and intersections, and determines whether control measures (such as starting emergency lanes or limiting the vehicle speed) need to be taken; if the monitoring equipment collects that the traffic flow of a certain high-speed road section suddenly increases, the rear-end control system can analyze whether the road section is congested according to prediction, and if the prediction result is congested, control measures for starting an emergency lane are adopted. The control instruction can be issued in a specific mode (such as a signpost, an LED drive-by-wire screen, an electronic police and a display screen) so as to enable a driver to adjust a driving strategy in time and ensure smooth operation of the expressway; the expressway management center acquires control condition feedback in real time through the mobile phone app, and modifies and perfects the control scheme according to feedback information, so that long-acting and stable operation of the system is ensured. The traffic flow fluctuation is dealt with in real time, the road traffic capacity is improved, and the congestion rate is reduced; real-time and accurate traffic information is provided for a driver, and the road use efficiency of the driver is improved; the personalized traffic control scheme is provided, intelligent management of the expressway is realized, and manpower and material resources are saved.

The beneficial effects of the technical scheme are as follows: the road traffic capacity is optimized, the congestion is relieved, and the overall traffic efficiency is improved; road safety and travel experience of drivers are improved; the intelligent and automatic traffic control is realized by using an advanced technology, and errors and risks caused by human intervention are reduced; the management and control scheme is timely adjusted and perfected, the actual traffic demand is met, and more effective traffic service is provided for users.

In another embodiment, frequent item set mining of ETC portal datasets obtains all frequently occurring item sets, including:

setting a threshold value of frequent item set mining according to experience or domain knowledge; generating all possible item sets from the ETC portal data set, and obtaining candidate item sets through combination and pruning; scanning the data set, counting the occurrence times of each candidate item set, and generating a frequent item set; generating an association rule based on the frequent item set, wherein the association rule comprises a front piece and a back piece, and the front piece and the back piece are item sets;

The working principle of the technical scheme is as follows: frequent item set mining and association rule generation methods aim to discover item-to-item associations from a large amount of ETC portal data.

Firstly, setting a proper support threshold according to experience or domain knowledge, ensuring that the mined frequent item set has practical significance, for example, when analyzing driver behavior data, selecting a lower support threshold to mine a relatively rare but characteristic behavior mode; in the commodity recommendation scene of the high-speed service station, a higher support threshold value can be selected to find common shopping combinations;

then, all possible item sets are generated from the ETC portal data set, and candidate item sets are obtained through combination and pruning, so that unnecessary calculation can be reduced in the pruning process, and the calculation efficiency is improved. For example, suppose that a dataset contains four items in total: A. b, C and D, firstly generating a candidate item set containing single items, then obtaining a binary item set by calculating the binary item set, and removing the item set with the support degree lower than a threshold value, if the support degree of the item set { A, B } is lower than the threshold value, the combination containing A and B is not considered when the ternary item set is generated, so that the calculation complexity is reduced;

Then scanning the data set, counting the occurrence times of each candidate item set, and further generating a frequent item set, specifically traversing each transaction in the data set for each candidate item set, and judging whether the item set exists in the transaction or not; if the item set exists, accumulating a counter, and finally calculating the support degree of each item set, and reserving the item set with the support degree reaching a threshold as a frequent item set;

finally, generating association rules according to the frequent item sets; this requires that a suitable confidence threshold is set first to ensure that the mined association rules have sufficient confidence, then for each frequent item set, association rules are generated for which the front and back pieces are non-empty (e.g., { a, B = > { C }), and the confidence is calculated, rules for which the confidence does not reach the threshold are deleted, and furthermore, association rules may be mined recursively, e.g., for which the front piece is non-empty { a = > { B, C }, other association rules related to the item set { B, C } may continue to be found.

The beneficial effects of the technical scheme are as follows: through frequent item set mining and association rule generation, key fields implicit in a large amount of data and relationships among the key fields can be found, which is helpful for deep understanding of the data set and provides valuable information for further analysis; the generated association rule can be applied to a plurality of fields, such as traffic congestion analysis, service quality evaluation and the like, and is beneficial to improving service efficiency, optimizing resource allocation, improving user satisfaction and the like; by reasonably setting the support degree threshold value, the confidence degree threshold value and the lifting degree threshold value, the accuracy and the practicability of the mining result can be ensured, and therefore an effective solution is provided for various application scenes.

It will be apparent to those skilled in the art that various modifications and variations can be made to the present invention without departing from the spirit or scope of the invention. Thus, it is intended that the present invention also include such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.

Claims

1. The highway traffic flow prediction method based on ETC portal data is characterized by comprising the following steps of:

inputting the preprocessed ETC portal key data into a trained multidimensional deep recurrent neural network, and predicting future traffic flow by utilizing the change rule of traffic flow learned by a model in multiple dimensions;

s103: based on the traffic flow prediction result, a traffic control scheme is formulated, traffic is managed according to the traffic control scheme, and the management measures comprise intelligent traffic scheduling, automatic command and congestion control measures;

The step S101 comprises the following steps:

s1013: automatically identifying and extracting key fields from the preprocessed ETC portal data by using an association rule mining algorithm to obtain ETC portal key data, wherein the ETC portal key data comprises vehicle passing time, vehicle identification numbers, vehicle types, vehicle speeds and lane occupancy;

the step S102 comprises the following steps:

s1023: inputting current ETC portal key data into a traffic flow early warning model, outputting a traffic flow prediction result, wherein the traffic flow prediction result comprises traffic flow early warning signals and key factors causing flow mutation, and providing targeted decision advice for traffic management departments by analyzing the early warning signals and the key factors in real time, wherein the traffic flow early warning signals are set based on real-time traffic information of traffic flow, speed and lane occupation, and comprise normal flow, crowded flow, serious congestion and extreme congestion, and the key factors causing flow mutation comprise traffic accidents, road construction and bad weather.

2. The method for predicting traffic flow of an expressway based on ETC portal data as recited in claim 1, wherein the step S103 comprises:

s1033: the method comprises the steps of carrying out real-time monitoring and evaluation on the effect of implementing management measures and traffic control schemes through big data and machine learning technology, automatically adjusting a control strategy according to an evaluation result and intelligently predicting future traffic data to realize continuous optimization of a traffic flow early warning model, wherein the big data technology is used for storing and processing the real-time traffic data through a distributed computing framework, and the machine learning technology is used for modeling the traffic flow prediction through a time sequence analysis model.

3. The method for predicting traffic flow of an expressway based on ETC portal data as recited in claim 1, wherein the step S1011 includes:

inputting the vehicle identification information into an ETC portal data generation algorithm model, acquiring ETC portal data comprising the vehicle identification information, and storing the ETC portal data into an ETC portal data management library.

4. The method for predicting traffic flow of an expressway based on ETC portal data as recited in claim 1, wherein step S1013 comprises:

5. The method for predicting traffic flow of an expressway based on ETC portal data as recited in claim 1, wherein step S1022 comprises:

6. The method for predicting traffic on an expressway based on ETC portal data according to claim 2, wherein S1031 includes:

in the process of adopting an intelligent optimization algorithm to formulate a traffic control scheme according to a traffic flow prediction result and a real traffic condition, a traffic control scheme is represented by chromosome codes, and the traffic control scheme comprises a specific control period, control measures and control points; initializing a population, and randomly generating a preset number of individuals, wherein the individuals are traffic control schemes; calculating individual fitness, and evaluating according to individual control measures and objective functions, wherein the objective functions comprise the degree of congestion relief and traffic flow capacity; selecting individuals in the population, and selecting a preset number of individuals as parents; performing genetic manipulation on the parent, including crossover and mutation, to produce offspring individuals; calculating the fitness of the offspring individuals, eliminating, and reserving the individuals with the best fitness to replace some individuals in the population; judging whether the condition of stopping iteration is met, and if the preset iteration times or the adaptability meets the preset requirement, outputting a traffic control scheme with the best final adaptability.

7. The method for predicting traffic flow of highway based on ETC portal data according to claim 2, wherein the real-time cross-regional coordinated monitoring and back-end control system in step S1032 automatically controls the high-speed access and opening the control measures of the emergency lane, comprising:

8. The method for predicting traffic flow of an expressway based on ETC portal data of claim 4, wherein frequent item set mining of the ETC portal data set is performed to obtain all frequently occurring item sets, comprising: