A method and system for traffic management based on traffic conditions

Abstract

This invention belongs to the field of traffic management and relates to a method and system for traffic management based on traffic conditions, comprising: Step 1, collecting multi-dimensional traffic data and preprocessing the traffic data to obtain preprocessed traffic data; Step 2, processing historical traffic data and current traffic data in the preprocessed traffic data to obtain predicted traffic conditions; Step 3, matching traffic management strategies from a traffic management strategy library based on the current traffic conditions and predicted traffic conditions; Step 4, implementing the traffic management strategies to achieve refined and intelligent control of the traffic system.

Description

Technical Field

[0001] This invention relates to the field of traffic management, and specifically discloses a method and system for traffic management based on traffic conditions. Background Technology

[0002] With the acceleration of urbanization and the surge in motor vehicle ownership, problems such as road traffic congestion and frequent traffic accidents have become increasingly prominent. Efficient traffic management has become crucial for improving the operational efficiency of the transportation system and ensuring traffic safety. Existing traffic management methods mostly rely on human experience or single-dimensional data for decision-making, resulting in problems such as delayed response and insufficient control precision, making it difficult to adapt to complex and ever-changing traffic situations.

[0003] In view of this, the present invention provides a method and system for traffic management based on traffic conditions, which integrates multi-dimensional traffic data, accurately assesses and predicts traffic conditions, and dynamically generates and optimizes management strategies to achieve refined and intelligent control of the traffic system. Summary of the Invention

[0004] The purpose of this invention is to provide a method for traffic management based on traffic conditions, addressing the problem of: achieving refined and intelligent control of the traffic system and improving control accuracy; the specific solution includes:

[0005] Step 1: Collect multi-dimensional traffic data and preprocess the traffic data to obtain preprocessed traffic data; Step 2: Process the historical traffic data and current traffic data in the preprocessed traffic data to obtain the predicted traffic situation; Step 3: Based on the current traffic situation and the predicted traffic situation, match traffic management strategies from the traffic management strategy library; Step 4: Implement the traffic management strategy.

[0006] Furthermore, step 2 includes: step 2.1, constructing a traffic situation assessment index system based on situation assessment indicators; step 2.2, constructing a traffic situation assessment model based on preprocessed traffic data and the traffic situation assessment index system; step 2.3, assessing the current traffic situation using the traffic situation assessment model to obtain the current traffic situation level; step 2.4, constructing a traffic situation prediction model based on historical traffic data and the current traffic situation level; and step 2.5, obtaining predicted traffic situations for multiple future time periods using the traffic situation prediction model.

[0007] Furthermore, the situation assessment indicators include traffic flow saturation, average speed deviation, and traffic incident impact index; the traffic flow saturation is determined based on the ratio of actual traffic flow to the maximum designed traffic flow of the road segment; the average speed deviation is determined based on the ratio of the difference between the actual average speed and the designed average speed of the road segment to the designed average speed of the road segment; and the traffic incident impact index is determined based on the type of incident and the scope of its impact.

[0008] Furthermore, step 3 includes: step 3.1, constructing a traffic management strategy library based on different traffic situation levels and traffic scenarios; step 3.2, matching an initial traffic management strategy from the traffic management strategy library based on the current traffic situation level and the predicted traffic situation; and step 3.3, optimizing the initial traffic management strategy using an optimization algorithm by combining real-time traffic data and road segment characteristics to obtain the final traffic management strategy.

[0009] Furthermore, it also includes: collecting traffic data after the implementation of traffic management strategies, and calculating traffic situation improvement indicators based on the traffic data; evaluating the implementation effect of traffic management strategies based on traffic situation improvement indicators to obtain traffic management strategy evaluation results; and updating traffic management strategies again based on traffic management strategy evaluation results.

[0010] This invention also provides a traffic management system based on traffic conditions, including a traffic data acquisition module, a traffic condition prediction module, a traffic management strategy determination module, and a traffic management implementation module. The traffic data acquisition module collects multi-dimensional traffic data and preprocesses the traffic data to obtain preprocessed traffic data. The traffic condition prediction module processes historical traffic data and current traffic data from the preprocessed traffic data to obtain predicted traffic conditions. The traffic management strategy determination module matches traffic management strategies from a traffic management strategy library based on the current traffic conditions and predicted traffic conditions. The traffic management implementation module implements the traffic management strategies.

[0011] Furthermore, the traffic situation prediction module includes an evaluation index system construction unit, a traffic situation evaluation model construction unit, a traffic situation level determination unit, a traffic situation prediction model construction unit, and a traffic situation prediction determination unit. The evaluation index system construction unit constructs a traffic situation evaluation index system based on situation evaluation indicators. The traffic situation evaluation model construction unit constructs a traffic situation evaluation model based on preprocessed traffic data and the traffic situation evaluation index system. The traffic situation level determination unit evaluates the current traffic situation using the traffic situation evaluation model to obtain the current traffic situation level. The traffic situation prediction model construction unit constructs a traffic situation prediction model based on historical traffic data and the current traffic situation level. The traffic situation prediction determination unit obtains predicted traffic situations for multiple future time periods using the traffic situation prediction model.

[0012] Furthermore, the situation assessment indicators include traffic flow saturation, average speed deviation, and traffic incident impact index; the traffic flow saturation is determined based on the ratio of actual traffic flow to the maximum designed traffic flow of the road segment; the average speed deviation is determined based on the ratio of the difference between the actual average speed and the designed average speed of the road segment to the designed average speed of the road segment; and the traffic incident impact index is determined based on the type of incident and the scope of its impact.

[0013] Furthermore, the traffic management strategy determination module includes a traffic management strategy library construction unit, an initial traffic management strategy matching unit, and a traffic management strategy optimization unit. The traffic management strategy library construction unit constructs a traffic management strategy library based on different traffic situation levels and traffic scenarios. The initial traffic management strategy matching unit matches an initial traffic management strategy from the traffic management strategy library based on the current traffic situation level and the predicted traffic situation. The traffic management strategy optimization unit combines real-time traffic data and road segment characteristics, and uses an optimization algorithm to optimize the initial traffic management strategy to obtain the final traffic management strategy.

[0014] Furthermore, it also includes a traffic situation improvement index calculation module, a traffic management strategy evaluation result determination module, and a traffic management strategy update module; the traffic situation improvement index calculation module collects the implemented traffic data after the implementation of the traffic management strategy and calculates the traffic situation improvement index based on the implemented traffic data; the traffic management strategy evaluation result determination module evaluates the implementation effect of the traffic management strategy based on the traffic situation improvement index and obtains the traffic management strategy evaluation result; the traffic management strategy update module updates the traffic management strategy again based on the traffic management strategy evaluation result.

[0015] The present invention has the following advantages and beneficial effects:

[0016] This invention solves the problems of single and one-sided data in the prior art by collecting and preprocessing traffic data from multiple dimensions, thereby improving the comprehensiveness and accuracy of traffic situation awareness and providing reliable data for subsequent analysis.

[0017] This invention achieves accurate classification of the current situation and effective prediction of the situation in multiple future time periods by constructing a systematic evaluation index system and prediction model.

[0018] This invention is based on a dynamic matching strategy of current and predicted situation, and optimizes the strategy by combining real-time data and road segment characteristics with an optimization algorithm. This solves the problem of existing preset fixed strategies being out of touch with actual needs, and improves the pertinence and flexibility of traffic management. Attached Figure Description

[0019] Figure 1 This is an exemplary flowchart of a traffic management method based on traffic conditions according to the present invention;

[0020] Figure 2 A data visualization of the acquired traffic data;

[0021] Figure 3 This is another screenshot of the traffic management strategy interface. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0023] Figure 1 This is an exemplary flowchart of a traffic management method based on traffic conditions according to the present invention. Figure 1 As shown, traffic management methods based on traffic conditions include:

[0024] Step 1: Collect multi-dimensional traffic data and preprocess it to obtain preprocessed traffic data. Traffic data refers to various data related to traffic. Multi-dimensional traffic data refers to multiple types of traffic data. Multi-dimensional traffic data can include traffic flow data, traffic event data, and meteorological data. Traffic flow data can include real-time traffic flow, average traffic speed, lane occupancy rate, real-time traffic volume, average speed, saturation, and congestion status. Traffic event data can include information on traffic accidents and road construction. Meteorological data can include information on rainfall and visibility. Preprocessing refers to cleaning the collected multi-dimensional traffic data, removing outliers and noise, filling in missing data, and standardizing the data. Preprocessed traffic data refers to the traffic data obtained after preprocessing the multi-dimensional traffic data. For example, by accessing traffic flow data such as traffic flow detection and video detection, real-time query and browsing of traffic operation information across the entire route can be achieved, including real-time traffic flow, average speed, saturation, and congestion status. All query results can be displayed in the form of lists or GIS maps.

[0025] Step 2 involves processing the historical and current traffic data in the preprocessed traffic data to obtain the predicted traffic situation. Historical traffic data refers to various traffic-related data from the past. For example, it includes comprehensive statistics and comparative analysis of traffic flow information across the entire route, including comprehensive statistics and comparative analysis of highway traffic flow information across different time dimensions (year, quarter, month, week, day, hour, etc.). Traffic volume statistics: Traffic flow data is collected based on vehicle inspection equipment, including the total and average traffic flow values ​​for specific equipment and time dimensions. Data can be statistically analyzed by year, month, and day. Real-time traffic volume data: Real-time data includes the latest reported traffic flow, average speed, and occupancy rate, calculated by vehicle inspection equipment and driving direction. Current traffic data refers to various data related to traffic information at the current time. Traffic situation refers to the traffic operation status and its development trend formed by the interaction of traffic flow (the flow, speed, density, etc. of moving objects such as vehicles and pedestrians), traffic events (traffic accidents, road construction, large-scale events, etc.), and the traffic environment (weather conditions, road conditions, traffic facility layout, etc.). Traffic situation forecasting refers to predicting future traffic conditions. Examples include traffic flow, speed, and congestion levels for the next 1 hour, 3 hours, and / or 24 hours.

[0026] In some embodiments, step 2 includes:

[0027] Step 2.1: Construct a traffic situation assessment index system based on situation assessment indicators. Situation assessment indicators are parameters used to quantify and describe the operational status of the traffic system. These indicators may include traffic flow saturation, average speed deviation, and traffic event impact index. Traffic flow saturation measures the degree of road congestion; a ratio closer to 1 indicates that traffic is closer to saturation and the road is more congested, while a smaller ratio indicates smoother traffic flow. Average speed deviation reflects the degree to which road speed is affected by traffic conditions; a larger deviation indicates a stronger inhibitory effect of current traffic conditions on speed and lower road efficiency. The traffic event impact index represents the impact of a traffic event on various indicators such as traffic flow, speed, and delays; a higher index value indicates a greater negative impact of the traffic event on overall traffic operation. Traffic flow saturation is determined by the ratio of actual traffic flow to the maximum designed traffic flow for a road segment; average speed deviation is determined by the ratio of the difference between the actual average speed and the designed average speed for the road segment to the designed average speed for the road segment; traffic incident impact index is determined based on the incident type and scope of impact. The traffic incident impact index can be calculated using mathematical models or algorithms based on the incident type and scope of impact from historical traffic incidents.

[0028] Step 2.2: Based on preprocessed traffic data and a traffic situation assessment index system, a traffic situation assessment model is constructed. The traffic situation assessment model refers to a neural network incorporating an attention mechanism, used to assess the traffic situation and obtain its level. The traffic situation level reflects the degree of traffic congestion and can include multiple levels. For example, the traffic situation level can be divided into three levels: level 3 represents smooth traffic, level 2 represents mild traffic congestion, and level 1 represents severe traffic congestion. A neural network architecture based on the fusion of multi-head attention and long short-term memory (LSTM) networks is used to construct the traffic situation assessment model. The model structure includes a feature extraction layer, an attention mechanism layer, an LSTM layer, and an output layer. The feature extraction layer uses a convolutional neural network (CNN) to extract spatial features; the attention mechanism layer automatically assigns weights to data from different time periods and road segments through a multi-head attention mechanism, focusing on key traffic features; the LSTM layer captures the temporal dependence and long-term dynamic changes of traffic flow; the output layer uses a fully connected layer to output the traffic situation assessment results at three levels. The traffic situation assessment model is obtained through model training. The model training uses labeled historical traffic data as the training set. By minimizing the mean squared error (MSE) or cross-entropy loss function, the Adam optimizer is used to iteratively update the model parameters to obtain the final traffic situation assessment model.

[0029] Step 2.3: Assess the current traffic situation using a traffic situation assessment model to obtain the current traffic situation level. The current traffic situation reflects the degree of traffic congestion at the current time. The current traffic situation can be obtained through real-time data, such as road traffic flow at the current time.

[0030] Step 2.4: Construct a traffic situation prediction model based on historical traffic data and the current traffic situation level. The traffic situation prediction model refers to an LSTM (Long Short-Term Memory) model that predicts traffic conditions for a future time period based on historical traffic data and real-time traffic conditions. The traffic situation prediction model uses LSTM as its core structure and consists of an input layer, multiple LSTM layers, a fully connected layer, and an output layer. The input layer receives preprocessed historical traffic data and real-time traffic situation information; the LSTM layers capture long-term dependencies and complex dynamic features of the traffic data through a gating mechanism; the fully connected layer integrates the features output by the LSTM layers; and the output layer outputs the predicted traffic situation for multiple future time periods. The traffic situation prediction model can be obtained through various feasible model training methods.

[0031] Step 2.5: Obtain the predicted traffic situation for multiple time periods in the future through the traffic situation prediction model.

[0032] Step 3: Based on the current and predicted traffic situation, traffic management strategies are matched from the traffic management strategy library. The traffic management strategy library is a structured collection of strategies, categorized and stored according to different traffic scenarios, traffic elements, and management objectives. Traffic scenarios include peak-hour congestion, accident handling, and large-scale events. Traffic elements include traffic flow, vehicle speed, and occupancy rate. Management objectives include traffic management efficiency, safety improvement, and environmental optimization. The construction of the strategy library is mainly achieved through the following steps: collecting traffic management cases from different cities and time periods, analyzing their applicable scenarios and implementation effects; using data mining and machine learning algorithms to extract strategy patterns from traffic data; optimizing and standardizing the strategies by incorporating expert opinions in the traffic field; and establishing a dynamic update mechanism to supplement and adjust the content of the strategy library based on new traffic characteristics. Traffic management strategies refer to a series of management measures and action plans for traffic management. Differentiated traffic management strategies emphasize developing targeted management plans based on different traffic situations, road types, regional functions, and other characteristics to achieve precise allocation and efficient control of traffic resources. For example, traffic management strategies can include congestion mitigation strategies and traffic flow management strategies. Congestion mitigation strategies refer to using variable lane control to dynamically adjust lane flow in areas experiencing sudden congestion; deploying mobile traffic signal optimization equipment to shorten idle phase time and improve intersection efficiency. Traffic flow management strategies refer to issuing guidance information in advance to divert vehicles based on predicted tidal traffic flow; opening backup roads or emergency lanes before peak hours to increase road network capacity; and implementing time-based traffic restrictions in key areas to limit the passage of high-polluting vehicles.

[0033] In some embodiments, step 3 includes:

[0034] Step 3.1: Construct a traffic management strategy library based on different traffic situation levels and traffic scenarios. Traffic scenarios can include holidays, severe weather, and emergencies. The traffic management strategy library can include various strategies such as diversion guidance, speed limit control, and emergency response. Fuzzy logic algorithms can be used to process traffic flow data, accident data, weather data, and holiday data to obtain the traffic situation levels.

[0035] ;

[0036] in, The traffic situation level is determined by the classification. , , and The first, second, third, and fourth weights are used to measure the importance of traffic flow data, accident data, weather data, and holiday data, respectively. These weights can be set based on experience, for example, all of them can be set to 1. , , and These represent traffic flow data, accident data, weather data, and holiday calendar data at time t, respectively; t is a time variable.

[0037] For each route, the weight of each backup route is dynamically calculated based on the sum of the route length from the current location to the destination, the estimated travel time, and the travel cost. The routes are then diverted and guided in descending order of weight.

[0038] Speed ​​limits are calculated based on traffic condition levels and real-time vehicle speed data, and the current speed limits are displayed via road LED screens and navigation systems.

[0039] ;

[0040] in, Indicates the speed limit; The speed limit is set as a benchmark for roads.

[0041] Emergency response strategies are implemented based on accident data and traffic conditions, including: activating the emergency response mechanism when accident data is greater than 0 and the traffic condition level is greater than or equal to 2. The emergency response mechanism includes setting up barriers upstream and downstream of the accident site and sending accident warnings; deploying the nearest tow truck based on the distance between the tow truck's location and the accident site; and adjusting traffic light timings at surrounding intersections in real time to create green channels for rescue vehicles. For example, in the case of severe congestion, diversion and guidance strategies are developed to guide vehicles to avoid congested sections; and emergency response strategies are developed for traffic accidents, including on-site rescue and traffic management.

[0042] Step 3.2: Based on the current traffic situation level and the predicted traffic situation, match an initial traffic management strategy from the traffic management strategy library. Based on the current traffic situation level assessment and the prediction results of future traffic conditions, select a suitable initial management strategy from the existing strategy library. For example, taking the morning rush hour in the main urban area of ​​a city as an example, the current traffic situation level is severe congestion (red alert), and it is predicted that the congestion area will spread along the main roads within the next hour. The system then matches strategies in the strategy library and selects multiple suitable strategies.

[0043] Step 3.3: Combining real-time traffic data and road segment characteristics, an optimization algorithm is used to optimize the initial traffic management strategy to obtain the final traffic management strategy, including:

[0044] Extract real-time traffic data and road segment features to obtain real-time traffic data. and road segment feature vector i represents the monitoring point variable; j represents the time step variable; and k represents the data type variable. For the i-th monitoring point, the k-th type of data is generated within the j-th time step; n represents the total number of road segment attributes. This represents the attribute of the nth road segment. Road segment attributes can include the number of lanes, speed limit, and intersection type, etc.

[0045] Real-time traffic data and road segment feature vectors are input into a spatiotemporal convolutional neural network to extract data features, resulting in a feature matrix F that integrates spatiotemporal information.

[0046] The initial traffic management strategy is evaluated and optimized using a deep Q-network (DQN) to generate an initial optimized traffic management strategy.

[0047] ;

[0048] in, This indicates the initial optimized traffic management strategy; PL represents the operation of selecting an initial traffic management strategy from the strategy library that matches the state space S; A represents the action space, which is the set of all operations that the traffic management system can perform, such as traffic light timing adjustment, lane function switching, variable speed limit control, and guidance information dissemination; S represents the state space, which is the joint vector of the feature matrix and the traffic management strategy evaluation results obtained after the execution of historical traffic management strategies; DQN represents a deep Q-network; PL represents the traffic management strategy library.

[0049] The initial optimized traffic management strategy is encoded as a chromosome, with each gene location corresponding to an action parameter. The chromosomes are repeatedly filtered and updated to obtain the final traffic management strategy. The fitness function of the genetic algorithm is:

[0050] ;

[0051] in, This represents the fitness function value of chromosome C; This represents the total time step; Let be the average vehicle speed at time t; Let be the traffic flow at time t; t represents the time variable.

[0052] Step 4: Implement traffic management strategies.

[0053] In some embodiments, it also includes:

[0054] Traffic data is collected after the implementation of traffic management strategies, and traffic situation improvement indicators are calculated based on this data. Traffic situation improvement indicators are quantitative standards used to assess the degree of optimization of the traffic system's operational status after the implementation of traffic management strategies. These indicators may include the rate of reduction in congestion duration and the rate of reduction in traffic accident occurrence.

[0055] Based on traffic situation improvement indicators, the effectiveness of traffic management strategies is evaluated, resulting in a traffic management strategy evaluation result. A multi-dimensional evaluation system is constructed based on traffic situation improvement indicators such as the rate of congestion duration reduction and the rate of traffic accident reduction. By comparing real-time traffic data with historical benchmark data, the effectiveness of traffic management strategies in alleviating congestion and reducing accident rates is quantitatively evaluated, forming the traffic management strategy evaluation result.

[0056] The traffic management strategy will be updated again based on the evaluation results, for subsequent fine-tuning of the traffic management strategy, including:

[0057] A quaternion is constructed based on the current traffic situation level, the final traffic management strategy, the evaluation results of the traffic management strategy, and the implementation of the traffic situation level.

[0058] The quadruple is input into a dual-network structure, and the policy parameters are updated using a near-end policy optimization algorithm. This further optimizes the final traffic management policy, resulting in another update to obtain the final traffic management policy. The dual-network structure includes an evaluation network. and target network The loss function for the dual-network structure is:

[0059] ;

[0060] in, Represents the loss function; This represents the expected value operation; p represents the evaluation result of the traffic management strategy. This is the discount factor, with a value range of [0,1]. To implement all possible traffic management strategies based on traffic situation levels The value is taken as the maximum value; The Q-value represents the traffic management strategy executed by the target network under the implemented traffic situation level. The target network is used to stabilize the learning process and reduce fluctuations during the training of the evaluation network. To evaluate the Q value of the network output for executing the final traffic management strategy under the current traffic situation level, the evaluation network evaluates the value of the action based on the current evaluation parameter θ. θ represents the network parameters of the target network; θ represents the network parameters of the evaluation network. The Q value represents the value of the final traffic management strategy output by the evaluation network after processing the current traffic situation level and the final traffic management strategy based on the current evaluation parameters θ. The value represents the target network based on the current target parameters. The value of the output traffic management strategy is derived from processing the input traffic situation level and traffic management strategy.

[0061] This invention also provides a system for traffic management based on traffic conditions, including a traffic data acquisition module, a traffic condition prediction module, a traffic management strategy determination module, and a traffic management implementation module. The traffic data acquisition module collects multi-dimensional traffic data and preprocesses the traffic data to obtain preprocessed traffic data. The multi-dimensional traffic data may include traffic flow data, traffic event data, and meteorological data, etc. Figure 2 As shown in the system interface, multi-dimensional traffic data can be acquired through various sensors as reporting devices. These sensors may include G7 weather detectors, energy consumption testing equipment, and vehicle detectors. The collected traffic data also includes the road segment, hazard type, and station number. The traffic situation prediction module processes historical and current traffic data from the preprocessed traffic data to obtain a predicted traffic situation. The traffic management strategy determination module matches the current and predicted traffic situations from a traffic management strategy library to obtain a traffic management strategy; the traffic management implementation module implements the traffic management strategy. Figure 3 As shown, traffic management strategies can be implemented by displaying guidance information to vehicles.

[0062] The traffic situation prediction module includes an evaluation index system construction unit, a traffic situation evaluation model construction unit, a traffic situation level determination unit, a traffic situation prediction model construction unit, and a traffic situation prediction determination unit. The evaluation index system construction unit constructs a traffic situation evaluation index system based on situation evaluation indicators. The traffic situation evaluation model construction unit constructs a traffic situation evaluation model based on preprocessed traffic data and the traffic situation evaluation index system. The traffic situation level determination unit evaluates the current traffic situation using the traffic situation evaluation model to obtain the current traffic situation level. The traffic situation prediction model construction unit constructs a traffic situation prediction model based on historical traffic data and the current traffic situation level. The traffic situation prediction determination unit obtains predicted traffic situations for multiple future time periods using the traffic situation prediction model. Situation evaluation indicators include traffic flow saturation, average speed deviation, and traffic event impact index. The traffic flow saturation is determined based on the ratio of actual traffic flow to the maximum designed traffic flow of the road segment. The average speed deviation is determined based on the ratio of the difference between the actual average speed and the designed average speed of the road segment to the designed average speed of the road segment. The traffic event impact index is determined based on the event type and scope of impact.

[0063] The traffic management strategy determination module includes a traffic management strategy library construction unit, an initial traffic management strategy matching unit, and a traffic management strategy optimization unit. The traffic management strategy library construction unit constructs a traffic management strategy library based on different traffic situation levels and traffic scenarios. The initial traffic management strategy matching unit matches an initial traffic management strategy from the traffic management strategy library based on the current traffic situation level and the predicted traffic situation. The traffic management strategy optimization unit combines real-time traffic data and road segment characteristics, and uses an optimization algorithm to optimize the initial traffic management strategy to obtain the final traffic management strategy.

[0064] In some embodiments, the system further includes a traffic situation improvement index calculation module, a traffic management strategy evaluation result determination module, and a traffic management strategy update module. The traffic situation improvement index calculation module collects traffic data after the implementation of the traffic management strategy and calculates the traffic situation improvement index based on the traffic data. The traffic management strategy evaluation result determination module evaluates the implementation effect of the traffic management strategy based on the traffic situation improvement index to obtain the traffic management strategy evaluation result. The traffic management strategy update module updates the traffic management strategy again based on the traffic management strategy evaluation result.

[0065] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for traffic management based on traffic situation, characterized by, The application relates to a traffic management strategy optimization method and device. The method comprises the following steps: Step 1, collecting multi-dimensional traffic data, and preprocessing the traffic data to obtain preprocessed traffic data; Step 2, processing historical traffic data and current traffic data in the preprocessed traffic data to obtain a predicted traffic situation; Step 3, matching a traffic management strategy from a traffic management strategy library based on the current traffic situation and the predicted traffic situation; the step 3 comprises: ; wherein, represents a speed limit speed; is a road reference speed limit; is a classified traffic situation level; Step 3.1, constructing a traffic management strategy library according to different traffic situation levels and traffic scenes; the traffic management strategy library comprises shunting induction, speed limiting control and emergency disposal; a speed limiting speed is calculated based on a traffic situation level and real-time vehicle speed data: Step 3.2, matching an initial traffic management strategy from the traffic management strategy library according to the current traffic situation level and the predicted traffic situation; Step 3.3, combining real-time traffic data and road section characteristics, and using an optimization algorithm to optimize the initial traffic management strategy to obtain a final traffic management strategy, which comprises: extracting real-time traffic data and road section characteristics to obtain real-time traffic data and road section characteristic vectors; inputting the real-time traffic data and road section characteristic vectors into a space-time convolutional neural network to extract data characteristics and obtain a feature matrix fused with space-time information; ; wherein, represents an initial optimized traffic management strategy; represents an operation of screening an initial traffic management strategy matching the state space S from the strategy library; A represents a set of all operations executable by the traffic management system; S represents a joint vector of the feature matrix and the traffic management strategy evaluation result obtained after the historical traffic management strategy execution; DQN represents a deep Q network; and PL represents a traffic management strategy library. using a deep Q network to evaluate and optimize the initial traffic management strategy to generate an initial optimized traffic management strategy: ; wherein, represents a fitness function value of chromosome C; is the total time step; is the average vehicle speed of the t period; is the traffic flow of the t period; t represents the time step variable; encoding the initial optimized traffic management strategy into a chromosome, with each gene site corresponding to an action parameter, repeatedly screening and updating the chromosome to obtain the final traffic management strategy; and the fitness function of the genetic algorithm is:

2. The method for traffic management based on traffic situation according to claim 1, wherein, Step 4, implementing the traffic management strategy. The step 2 comprises: Step 2.1, constructing a traffic situation evaluation index system based on a situation evaluation index; Step 2.2, constructing a traffic situation evaluation model based on the preprocessed traffic data and the traffic situation evaluation index system; Step 2.3, evaluating the current traffic situation through the traffic situation evaluation model to obtain a current traffic situation level; Step 2.4, constructing a traffic situation prediction model based on historical traffic data and the current traffic situation level; 3. The method for traffic management based on traffic situation according to claim 2, wherein, Step 2.5, obtaining predicted traffic situations of multiple future time periods through the traffic situation prediction model. The situation evaluation index comprises a traffic flow saturation degree, an average vehicle speed deviation degree and a traffic event influence index; the traffic flow saturation degree is determined according to the ratio of actual traffic flow to the maximum traffic flow of a road section design; The average vehicle speed deviation degree is determined according to the ratio of the difference between actual average vehicle speed and road section design average vehicle speed to the road section design average vehicle speed; 4. The method for traffic management based on traffic situation according to claim 1, wherein, The traffic event influence index is determined according to the event type and the influence range. The application further comprises: collecting implementation traffic data after the traffic management strategy is implemented, and calculating a traffic situation improvement index according to the implementation traffic data; evaluating the implementation effect of the traffic management strategy based on the traffic situation improvement index to obtain a traffic management strategy evaluation result; updating the traffic management strategy again based on the traffic management strategy evaluation result, which comprises: constructing a four-tuple based on the current traffic situation level, the final traffic management strategy, the traffic management strategy evaluation result and the implementation traffic situation level; The quadruplet is input into a double network structure, and a proximal policy optimization algorithm is used to update policy parameters, so as to optimize a final traffic management policy, and update the traffic management policy; the double network structure comprises an evaluation network and a target network ; and a loss function of the double network structure is: ; wherein, represents a loss function; represents an expectation operation; p represents the evaluation result of the traffic management strategy; is a discount factor; is the Q value of the action traffic management strategy performed under the traffic situation level for all possible traffic management strategies; takes the maximum value; represents the Q value of the action traffic management strategy performed under the traffic situation level for the target network output; is the Q value of the final traffic management strategy performed under the current traffic situation level for the evaluation network output; represents the network parameters of the target network; θ represents the network parameters of the evaluation network.

5. A system for traffic management based on traffic situation, characterized by, The traffic data collection module, the traffic situation prediction module, the traffic management strategy determination module and the traffic management implementation module are comprised; The traffic data collection module collects multi-dimensional traffic data and pre-processes the traffic data to obtain pre-processed traffic data; The traffic situation prediction module processes historical traffic data and current traffic data in the pre-processed traffic data to obtain a predicted traffic situation; The traffic management strategy determination module matches a traffic management strategy from a traffic management strategy library based on the current traffic situation and the predicted traffic situation; the traffic management strategy determination module comprises a traffic management strategy library construction unit, an initial traffic management strategy matching unit and a traffic management strategy optimization unit; The traffic management strategy library construction unit constructs a traffic management strategy library according to different traffic situation levels and traffic scenarios; the traffic management strategy library comprises diversion and induction, speed limit control and emergency disposal; a speed limit speed is calculated based on a traffic situation level and real-time vehicle speed data: ; wherein, represents a speed limit speed; is a road reference speed limit; is a classified traffic situation level; The initial traffic management strategy matching unit matches an initial traffic management strategy from the traffic management strategy library according to the current traffic situation level and the predicted traffic situation; The traffic management strategy optimization unit optimizes the initial traffic management strategy by combining real-time traffic data and road segment characteristics using an optimization algorithm to obtain a final traffic management strategy, comprising: extracting real-time traffic data and road segment characteristics to obtain real-time traffic data and road segment characteristic vectors; inputting the real-time traffic data and road segment characteristic vectors into a spatio-temporal convolutional neural network to extract data features and obtain a feature matrix fused with spatio-temporal information; evaluating and optimizing the initial traffic management strategy using a deep Q network to generate an initial optimized traffic management strategy: ; wherein, represents an initial optimized traffic management strategy; represents an operation of screening an initial traffic management strategy matching the state space S from the strategy library; A represents a set of all operations executable by the traffic management system; S represents a joint vector of the feature matrix and the traffic management strategy evaluation result obtained after the historical traffic management strategy execution; DQN represents a deep Q network; and PL represents a traffic management strategy library. encoding the initial optimized traffic management strategy as a chromosome, with each gene position corresponding to an action parameter, repeatedly screening and updating the chromosome to obtain a final traffic management strategy; the fitness function of the genetic algorithm is: ; wherein, represents a fitness function value of chromosome C; is the total time step; is the average vehicle speed of the t period; is the traffic flow of the t period; t represents the time step variable; The traffic management implementation module implements the traffic management strategy. 6.The system for traffic management based on traffic situation according to claim 5, wherein, The traffic situation prediction module comprises an evaluation index system construction unit, a traffic situation evaluation model construction unit, a traffic situation level determination unit, a traffic situation prediction model construction unit and a predicted traffic situation determination unit; The evaluation index system construction unit constructs a traffic situation evaluation index system based on a situation evaluation index; The traffic situation evaluation model construction unit constructs a traffic situation evaluation model based on pre-processed traffic data and a traffic situation evaluation index system; The traffic situation level determination unit evaluates the current traffic situation through the traffic situation evaluation model to obtain a current traffic situation level; The traffic situation prediction model construction unit constructs a traffic situation prediction model based on historical traffic data and the current traffic situation level; The predicted traffic situation determination unit obtains predicted traffic situations for multiple future time periods through the traffic situation prediction model. 7.The system for traffic management based on traffic situation according to claim 6, wherein, The situation evaluation index comprises traffic flow saturation, average vehicle speed deviation and traffic event influence index; the traffic flow saturation is determined according to the ratio of actual traffic flow to the maximum traffic flow designed for a road segment; The average vehicle speed deviation degree is determined according to a ratio of a difference between an actual average vehicle speed and a road section design average vehicle speed to the road section design average vehicle speed; The traffic event influence index is determined according to an event type and an influence range. 8.The system for traffic management based on traffic situation according to claim 5, wherein, The traffic situation improvement index calculation module, the traffic management strategy evaluation result determination module and the traffic management strategy updating module are further included; The traffic situation improvement index calculation module collects implementation traffic data after the traffic management strategy is implemented, and calculates a traffic situation improvement index according to the implementation traffic data; The traffic management strategy evaluation result determination module evaluates an implementation effect of the traffic management strategy based on the traffic situation improvement index, and obtains a traffic management strategy evaluation result; The traffic management strategy updating module updates the traffic management strategy again based on the traffic management strategy evaluation result, including: A quadruple is constructed based on a current traffic situation grade, a final traffic management strategy, a traffic management strategy evaluation result and an implementation traffic situation grade; The quadruplet is input into a double network structure, and a proximal policy optimization algorithm is used to update policy parameters, so as to optimize a final traffic management policy, and update the traffic management policy; the double network structure comprises an evaluation network and a target network ; and a loss function of the double network structure is: ; wherein, represents a loss function; represents an expectation operation; p represents the evaluation result of the traffic management strategy; is a discount factor; is the Q value of the target network output for performing the action traffic management strategy under the implementation of the traffic situation level; takes the maximum value; represents the Q value of the target network output for performing the action traffic management strategy under the implementation of the traffic situation level; is the Q value of the evaluation network output for performing the final traffic management strategy under the current traffic situation level; represents the network parameter of the target network; θ represents the network parameter of the evaluation network.

Images