Vehicle-road collaborative artificial intelligence traffic light optimization scheduling method and system

Abstract

The invention provides a vehicle-road cooperation artificial intelligence traffic light optimization scheduling method and system, and relates to the technical field of intelligent traffic, and the method comprises the steps: obtaining the road traffic data of a target road through connecting a vehicle-road cooperation system; performing feature combing according to the road traffic data, and determining road traffic feature information; traffic light scheduling demand identification is carried out according to the road traffic characteristic information to obtain traffic light scheduling demands; adjusting the traffic light scheduling real-time scheme according to traffic light scheduling requirements to obtain a traffic light scheduling adjustment scheme; and executing traffic light optimization scheduling of the target road according to the traffic light scheduling adjustment scheme. According to the method and the device, the technical problems of traffic jam and traffic flow waste caused by mismatching of signal lamp timing and actual road conditions due to the fact that dynamic changes of traffic flow cannot be responded in real time in the prior art are solved, and the technical effects of optimizing traffic resource allocation and further improving road traffic efficiency are achieved.

Description

Technical Field

[0001] This application relates to the technical field of intelligent transportation, and particularly to an artificial intelligence traffic light optimization scheduling method and system for vehicle-road cooperation. Background Art

[0002] Traffic light scheduling is a core link in urban traffic management, directly affecting traffic fluency, road safety, and drivers' travel experience.

[0003] Most existing traffic light scheduling methods are based on relatively fixed patterns, such as switching traffic light durations according to a preset schedule, or relying on data obtained from simple traffic flow counter sensor devices. Such traditional scheduling methods have many deficiencies. On the one hand, the fixed-time control method ignores the changes in real-time traffic flow, fails to comprehensively understand the traffic conditions on the road, and cannot accurately reflect the actual traffic pressure on the road. On the other hand, it lacks flexibility. In the face of dynamic changes in traffic flow, such as rush hours or sudden traffic incidents, it cannot adjust the traffic light duration in a timely manner, which easily leads to increased traffic congestion and ineffective utilization of road resources. Summary of the Invention

[0004] This application provides an artificial intelligence traffic light optimization scheduling method and system for vehicle-road cooperation, which solves the technical problem that the existing technology, due to relying on a static scheduling scheme, cannot respond to the dynamic changes of traffic flow in real time, resulting in a mismatch between signal timing and actual road conditions, thereby causing traffic congestion and waste of traffic flow, and achieves the technical effect of optimizing traffic resource allocation and then improving road traffic efficiency.

[0005] In view of the above problems, on the one hand, this application provides an artificial intelligence traffic light optimization scheduling method for vehicle-road cooperation. The method includes: connecting to a vehicle-road cooperation system to obtain road traffic data of a target road; sorting out features based on the road traffic data to determine road traffic feature information; identifying traffic light scheduling requirements based on the road traffic feature information to obtain traffic light scheduling requirements; obtaining a real-time traffic light scheduling plan for the target road; adjusting the real-time traffic light scheduling plan according to the traffic light scheduling requirements to obtain a traffic light scheduling adjustment plan; and performing optimized traffic light scheduling on the target road according to the traffic light scheduling adjustment plan.

[0006] On the other hand, the present application also provides an artificial intelligence traffic light optimization scheduling system for vehicle-road cooperation, which includes: a traffic data acquisition module for connecting to the vehicle-road cooperation system to obtain the road traffic data of the target road; a feature sorting module for sorting out features based on the road traffic data to determine the road traffic feature information; a scheduling requirement identification module for identifying the traffic light scheduling requirements based on the road traffic feature information to obtain the traffic light scheduling requirements; a scheme acquisition module for obtaining the real-time traffic light scheduling scheme of the target road; a scheduling adjustment module for adjusting the real-time traffic light scheduling scheme according to the traffic light scheduling requirements to obtain a traffic light scheduling adjustment scheme; and a scheduling execution module for performing the optimized scheduling of the traffic lights on the target road according to the traffic light scheduling adjustment scheme.

[0007] One or more technical solutions provided in the present application have at least the following technical effects or advantages:

[0008] By connecting to the vehicle-road cooperation system, it is possible to obtain the real-time traffic data of the target road, providing rich road traffic information for subsequent analysis and scheduling. By sorting out the features of the acquired traffic data, the key road traffic feature information is extracted, providing a basis for subsequent identification of traffic light scheduling requirements. Based on the extracted road traffic feature information, the specific requirements for traffic light scheduling are identified, providing a direction for subsequent formulation of the scheduling scheme. According to the identified traffic light scheduling requirements, the initially obtained traffic light scheduling scheme is adjusted specifically, making the traffic light scheduling scheme more suitable for the current traffic conditions and improving the accuracy of scheduling. According to the finally determined traffic light scheduling adjustment scheme, the traffic lights on the target road are optimized for scheduling, reasonably allocating the traffic flow and improving the road traffic efficiency.

[0009] In summary, by connecting to the vehicle-road cooperation system, the present application obtains comprehensive road traffic data, then conducts in-depth feature sorting and requirement identification, and then adjusts and executes the initial traffic light scheduling scheme, enabling dynamic adjustment of traffic light scheduling according to the actual traffic conditions, optimizing the allocation of traffic resources, thereby effectively improving the road traffic efficiency, reducing traffic congestion, enhancing the utilization rate of road resources, and strengthening the ability to respond to sudden traffic conditions.

[0010] The above description is only an overview of the technical solutions of the present application. In order to be able to understand the technical means of the present application more clearly, it can be implemented according to the content of the specification. And in order to make the above and other purposes, features and advantages of the present application more obvious and understandable, the following specifically illustrates the specific embodiments of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 It is a schematic flowchart of the artificial intelligence traffic light optimization scheduling method for vehicle-road cooperation provided by the embodiment of the present application.

[0012] Figure 2 This is a schematic structural diagram of the vehicle-road collaborative artificial intelligence traffic light optimization scheduling system provided by the embodiments of the present application.

[0013] Explanation of reference numerals: Traffic data acquisition module 10, feature sorting module 20, scheduling requirement identification module 30, solution acquisition module 40, scheduling adjustment module 50, scheduling execution module 60. Detailed implementation manners

[0014] By providing the vehicle-road collaborative artificial intelligence traffic light optimization scheduling method and system in the embodiments of the present application, the technical problem in the prior art that due to relying on a static scheduling scheme and being unable to respond to the dynamic changes of traffic flow in real time, the signal timing does not match the actual road conditions, resulting in traffic congestion and waste of traffic flow, is solved, and the technical effect of optimizing the allocation of traffic resources and thus improving the road traffic efficiency is achieved.

[0015] Embodiment 1, as Figure 1 shown, the embodiments of the present application provide a vehicle-road collaborative artificial intelligence traffic light optimization scheduling method, and the method includes:

[0016] Step S1: Connect to the vehicle-road collaborative system to obtain the road traffic data of the target road.

[0017] Specifically, the vehicle-road collaborative system is an intelligent transportation system integrating vehicles, road infrastructure, and communication technologies. Through the information interaction between vehicles and road infrastructure, real-time sharing and collaborative control of traffic information are realized. The vehicle-road collaborative system includes a variety of devices and technologies, such as intelligent sensors (traffic flow sensors, vehicle speed sensors, etc. installed on the road), communication devices (such as DSRC, 5G communication devices), computing units (for processing the collected information), etc. By connecting to the vehicle-road collaborative system through a communication interface (such as a 5G network interface or a DSRC interface), various sensors in the vehicle-road collaborative system (such as cameras, radar sensors, in-vehicle devices, etc.) will collect the traffic data on the target road, including real-time traffic flow, vehicle speed, traffic density, traffic events (such as accidents, construction), weather conditions, and other information. For example, the camera can capture image information such as the number, type, and queue length of vehicles on the road, and the radar sensor can obtain data such as the vehicle speed and distance. These data are transmitted to the data processing center through the network.

[0018] By connecting to the vehicle-road collaborative system in real time, accurate traffic data can be obtained, and real-time information can be provided for subsequent traffic scheduling decisions, enabling subsequent scheduling decisions to make dynamic responses based on current traffic flow, vehicle speed, and other information.

[0019] Step S2: Sort out the features according to the road traffic data to determine the road traffic feature information.

[0020] Specifically, the collected road traffic data is sorted, classified, and analyzed to identify meaningful attributes or patterns that can reflect the essence of the traffic conditions, which are used as road traffic characteristics. Road traffic characteristics include time characteristics, spatial characteristics, and dynamic characteristics of road traffic data. For example, time characteristics include peak hours, off-peak hours, low-peak hours, etc.; spatial characteristics include vehicle density, speed distribution, and queue length, etc.; dynamic characteristics include acceleration and deceleration patterns of vehicles, lane-changing frequency, etc. Feature sorting can use data analysis algorithms and software tools, such as the Pandas and Numpy libraries in Python. For traffic flow data, it can be classified and statistically analyzed according to different time periods (such as every 15 minutes as a time period) to find the peak and trough hours of traffic flow; for vehicle type data, the proportions of different types of vehicles (such as cars, trucks, buses, etc.) can be statistically analyzed.

[0021] Through feature sorting, key feature information can be extracted from a large amount of traffic data, providing accurate data support for the subsequent identification of traffic light scheduling requirements, so as to make more accurate scheduling decisions based on the real-time traffic conditions.

[0022] Step S3: Identify the traffic light scheduling requirements based on the road traffic characteristic information to obtain the traffic light scheduling requirements.

[0023] Specifically, using the road traffic characteristic information obtained in step S2, the specific requirements for traffic light scheduling under the current traffic conditions are judged through machine learning algorithms (such as neural network algorithms, decision tree algorithms, etc.) to determine the current traffic light scheduling requirements. For example, if the traffic flow in a certain direction continues to increase during peak hours, the green light time in that direction needs to be extended; if it is found that vehicles often queue up too long at a certain intersection, the switching frequency of the traffic lights needs to be adjusted.

[0024] Through the identification of traffic light scheduling requirements, the traffic light scheduling requirements can be dynamically adjusted according to the traffic conditions, providing clear guidance for the adjustment of traffic lights, improving the flexibility and adaptability of traffic light scheduling, helping to avoid traffic congestion, and improving road traffic efficiency.

[0025] Step S4: Obtain the real-time traffic light scheduling plan for the target road.

[0026] Specifically, the real-time traffic light scheduling plan is a preliminary plan for the traffic light time arrangement and switching logic of the target road, formulated by the traffic model or scheduling rules configured by the current traffic light scheduling system. Interact with the current traffic light scheduling system to obtain the real-time traffic light scheduling plan currently in use. This plan is the basis for subsequent optimization and adjustment, providing a basic framework for the entire scheduling optimization process.

[0027] Step S5: Adjust the real-time traffic light scheduling plan according to the traffic light scheduling requirements to obtain a traffic light scheduling adjustment plan.

[0028] Specifically, compare the scheduling requirements in Step S3 with the real-time plan in Step S4, and use an algorithm to adjust the real-time plan. For example, it is identified that a certain direction needs to extend the green light time. In the real-time plan, the green light duration in this direction is 30 seconds, and it is extended to 40 seconds according to the requirements. The algorithms used can be rule-based algorithms or intelligent optimization algorithms (such as genetic algorithms, etc.). By adjusting the real-time plan, the traffic light scheduling plan can better fit the current traffic conditions, be flexibly adjusted according to actual needs, effectively improve traffic flow, and reduce vehicle waiting time.

[0029] Step S6: Execute the optimized traffic light scheduling for the target road according to the traffic light scheduling adjustment plan.

[0030] Specifically, a traffic signal controller (a hardware device that can receive control instructions and control the lighting and duration of traffic lights) is used to execute the traffic light scheduling adjustment plan in Step S5. For example, parameters such as the green light duration and red light duration in the adjustment plan are sent to the traffic signal controller, and the traffic signal controller controls the operation of the traffic lights according to these parameters, ultimately realizing the optimized scheduling of the traffic lights on the target road, improving traffic flow, reducing traffic congestion, and enhancing the utilization rate of road resources.

[0031] Further, Step S2 of the embodiment of the present application includes:

[0032] Step S21: Preprocess the road traffic data to obtain standard road traffic data.

[0033] Step S22: Identify time features based on the standard road traffic data to obtain the first road traffic feature data.

[0034] Step S23: Identify spatial features based on the standard road traffic data to obtain the second road traffic feature data.

[0035] Step S24: Identify dynamic features based on the standard road traffic data to obtain the third road traffic feature data.

[0036] Step S25: Perform data fusion based on the first road traffic feature data, the second road traffic feature data, and the third road traffic feature data, and output the road traffic feature information.

[0037] Specifically, the original road traffic data may contain some outliers or error values (such as vehicle positions outside the road range). It is necessary to preprocess the received traffic data, including removing noise, filling in missing values, normalizing the data format, etc., to ensure that the data format is unified and noise-free, and obtain the standard road traffic data. The preprocessed standard data provides a higher-quality data foundation, ensuring that subsequent analysis and feature recognition steps can be more accurate and efficient, and avoiding incorrect analysis caused by data inconsistency or noise interference.

[0038] Analyze the preprocessed standard road traffic data according to the time dimension, extract time-related features, and obtain the first feature data of road traffic, including peak hours, flat peak hours, low peak hours of traffic flow in a day, and the variation law of traffic flow on different dates within a week, etc. For example, divide 24 hours of a day into certain time intervals (such as each hour as an interval), and count data such as vehicle flow and average vehicle speed within each interval. Use data analysis tools, such as the Pandas library in Python, to group and count the data according to time. By analyzing the data change trend in different time periods, determine the first feature data of road traffic.

[0039] Consider the spatial layout of the road, such as factors like the number of lanes on different road sections and the connection directions of intersections, analyze data such as vehicle flow and vehicle speed at different positions in the standard road traffic data, determine spatial-related features, such as traffic flow differences at different road sections and intersections, vehicle density, speed distribution, and queue length of each lane, etc., and obtain the second feature data of road traffic. Geographic Information System (GIS) tools, such as ArcGIS software, can be used to combine traffic data with the spatial information of the road. For example, it is found that the traffic flow is larger on the road sections near the commercial center and smaller on the road sections far from the urban area.

[0040] Analyze the vehicle movement trajectory data, find out the features reflecting traffic dynamic changes in the standard road traffic data, such as vehicle acceleration, deceleration, lane-changing frequency, etc., and the impact of these dynamic behaviors on traffic flow. For example, determine the acceleration and deceleration of vehicles by analyzing the vehicle speed change rate, and count the number of lane changes of vehicles within a certain time. Use traffic analysis software, such as the microscopic traffic analysis function in VISSIM, to simulate and analyze the dynamic behaviors of vehicles, so as to obtain the third feature data of road traffic, such as frequent lane changes on a certain road section are likely to cause traffic congestion, etc.

[0041] Using data fusion algorithms, such as the weighted average method or the fusion algorithm based on the theory of evidence, integrate the obtained first, second, and third characteristic data of road traffic to form an information set that comprehensively reflects the characteristics of road traffic, namely, road traffic characteristic information. Exemplarily, the weight of the first characteristic data of road traffic emphasizing the influence of time factors on traffic is 0.4, the weight of the second characteristic data of road traffic emphasizing spatial factors is 0.3, and the weight of the third characteristic data of road traffic emphasizing dynamic factors is 0.3. Perform weighted average fusion on the first, second, and third characteristic data according to this weight, and use a database management system (such as an Oracle database) to store and manage the fused data, and finally output the road traffic characteristic information.

[0042] By integrating road traffic characteristics from different dimensions, the output road traffic characteristic information can comprehensively and accurately reflect the traffic conditions, providing a sufficient basis for subsequent identification of traffic light scheduling requirements.

[0043] Further, step S21 in the embodiment of the present application includes:

[0044] Step S211: Perform data cleaning on the road traffic data according to a predetermined cleaning factor to obtain the first processing result of the traffic data.

[0045] Step S212: Perform Kalman filtering based on the first processing result of the traffic data to obtain the second processing result of the traffic data.

[0046] Step S213: Perform normalization processing based on the second processing result of the traffic data to generate the road traffic standard data.

[0047] Specifically, the preprocessing of road traffic data includes specific steps such as data cleaning, noise filtering, and normalization processing. First, process the original road traffic data according to predetermined cleaning rules (i.e., predetermined cleaning factors, including outlier correction, duplicate value deletion, and missing value filling, etc.) to improve the data quality, remove errors, redundancies, and incomplete information therein, and obtain the first processing result of the traffic data, providing a cleaner and more reliable data basis for subsequent further processing.

[0048] Taking the first processing result of traffic data as the input, based on the algorithm principle of Kalman filtering, establish the state equation and observation equation of traffic data to obtain the second processing result of traffic data. For example, the traffic flow in traffic data is a state quantity that changes over time. Through the Kalman filtering algorithm, the state estimate value is updated according to the previous state estimate value and the current observation value (such as the traffic flow data collected by sensors). In actual operation, tools such as MATLAB can be used to implement the Kalman filtering algorithm to filter traffic data. Kalman filtering can effectively reduce the noise interference in traffic data, make the data smoother, and better reflect the real change trend of traffic data.

[0049] Perform normalization processing on the second processing result of traffic data, convert the data to a specific interval range according to certain rules, usually the [0,1] interval or the [-1,1] interval, to generate road traffic standard data. After normalization processing, the road traffic standard data has a unified dimension, which is convenient for subsequent comprehensive analysis of road traffic data. For example, when performing operations such as data fusion, data with different characteristics can be compared and calculated on the same scale. In actual operation, the Numpy library in Python can be used to perform efficient normalization calculations.

[0050] Through data cleaning, Kalman filtering and smoothing processing, as well as normalization processing, more accurate and stable traffic information can be extracted from the original traffic data with more noise. This process eliminates the outliers and fluctuations in the data, making the finally output traffic standard data of higher quality, providing a reliable basis for subsequent feature recognition and data analysis, and significantly improving the usability of road traffic data and the accuracy of analysis.

[0051] Furthermore, the predetermined cleaning factors described in the embodiments of the present application include outlier correction, duplicate value deletion, and missing value filling.

[0052] Specifically, the predetermined cleaning factors in the data cleaning step include outlier correction, duplicate value deletion, and missing value filling. Outlier correction is to adjust the values that are significantly deviated from the normal range due to reasons such as sensor failures and data transmission errors to reasonable values or directly eliminate them; duplicate value deletion is to remove the data that is repeatedly collected due to reasons such as failures of data acquisition devices and program errors; missing value filling is to supplement the missing part of the data due to various reasons (such as sensor damage, data transmission interruption, etc.).

[0053] For outlier correction, a reasonable data range can be set first. For example, for traffic flow data, a normal fluctuation range can be determined based on historical data. When data is detected to be outside this range, various methods can be used for correction, such as replacing it with the average or median of adjacent data. For duplicate value deletion, a data duplication checking algorithm can be used to identify and delete duplicate data by comparing adjacent data or based on the unique identifier of the data (such as the collection timestamp, etc.). For missing value filling, a suitable method can be selected according to the type and characteristics of the data. If it is continuous data (such as vehicle speed), linear interpolation or filling based on other relevant data (such as the relationship between traffic flow and vehicle speed) can be used; if it is discrete data (such as vehicle type statistics), filling can be performed according to the proportion of the same type of data in other time periods or sections.

[0054] By preprocessing the road traffic data using a predetermined cleaning factor (outlier correction, duplicate value deletion, and missing value filling), the quality of the data can be improved, making subsequent analysis based on this data (such as time feature recognition, spatial feature recognition, and dynamic feature recognition, etc.) more accurate and reliable, thereby improving the effectiveness of the entire traffic light scheduling scheme.

[0055] Furthermore, step S3 of the embodiment of the present application includes:

[0056] Step S31: Obtain a road traffic feature sample set and a traffic light scheduling requirement sample set.

[0057] Step S32: Perform supervised learning based on the road traffic feature sample set and the traffic light scheduling requirement sample set to obtain a traffic light scheduling requirement parsing model.

[0058] Step S33: Input the road traffic feature information into the traffic light scheduling requirement parsing model and output the traffic light scheduling requirement.

[0059] Specifically, the road traffic feature sample set is a set of sample collections containing road traffic feature data. Each sample data includes road traffic features in multiple dimensions such as time, space, and dynamic features, reflecting the traffic state of the road under specific time periods or spatio-temporal conditions. The road traffic feature sample set can be extracted from historical traffic data, and these historical data can come from the databases of traffic management departments, which contain traffic monitoring data of various roads. For example, from the traffic monitoring data of the main roads in a certain city in the past year, the road traffic feature data in different time periods and different weather conditions are screened out to form the road traffic feature sample set.

[0060] The traffic light scheduling demand sample set is a set of traffic light scheduling demand samples related to traffic feature data. Each sample reflects the required traffic light signal timing demands under a specific traffic state, such as extending the green light duration or adjusting the switching frequency of traffic lights. The traffic light scheduling demand sample set needs to be sorted out in combination with historical traffic data and the actual traffic light scheduling plan implemented at that time. Data mining tools, such as Weka software, can be used to mine and sort out the historical data to obtain a sample set that meets the requirements.

[0061] Select a suitable supervised learning algorithm, such as the decision tree algorithm or the neural network algorithm, use the road traffic feature sample set as the input feature, and the traffic light scheduling demand sample set as the target value for supervised learning to obtain a traffic light scheduling demand parsing model. This model can accurately parse the corresponding traffic light scheduling demands according to the input road traffic features, providing an intelligent decision-making basis for traffic light scheduling. Input the road traffic feature information obtained in step S2 into the constructed traffic light scheduling demand parsing model, and the model will output the corresponding traffic light scheduling demands according to the internal decision rules.

[0062] The above steps train a traffic light scheduling demand parsing model through supervised learning, predict the corresponding traffic light scheduling demands based on real-time road traffic feature information, and thus can automatically adjust the traffic light timing accordingly. This process greatly improves the intelligence and automation of traffic light scheduling, can respond to changes in road traffic flow in real time, optimize the timing arrangement of traffic lights, and thus reduce traffic congestion and improve road traffic efficiency.

[0063] Further, step S32 of the embodiment of the present application includes:

[0064] Step S321: Perform data cleaning on the road traffic feature sample set and the traffic light scheduling demand sample set to obtain a traffic light scheduling demand record set.

[0065] Step S322: Perform data partitioning on the traffic light scheduling demand record set to obtain a traffic light scheduling demand training set and a traffic light scheduling demand test set.

[0066] Step S323: Train and test a fully connected neural network according to the traffic light scheduling demand training set and the traffic light scheduling demand test set to generate the traffic light scheduling demand parsing model.

[0067] Specifically, during the supervised learning process based on the road traffic feature sample set and the traffic light scheduling requirement sample set, first, data cleaning is performed on the road traffic feature sample set and the traffic light scheduling requirement sample set to remove error data, missing data, duplicate data, etc., and a traffic light scheduling requirement record set is obtained. Each record in this traffic light scheduling requirement record set includes a complete traffic feature input and the corresponding traffic light scheduling requirement output. The specific data cleaning process can refer to the relevant content of step S21 above.

[0068] The traffic light scheduling requirement record set after cleaning is divided into two parts according to a certain ratio. One part is used to train the fully connected neural network, denoted as the traffic light scheduling requirement training set, and the other part is used to test the trained network, denoted as the traffic light scheduling requirement test set. Common division ratios can be 80:20 or 70:30, etc.

[0069] Use a deep learning framework, such as TensorFlow or PyTorch, to build a fully connected neural network. For the training process, use the road traffic feature data of the traffic light scheduling requirement training set as the input and the traffic light scheduling requirement data as the output, and adjust the weights and biases of the network through the backpropagation algorithm to minimize the error between the prediction result and the actual result. For example, in TensorFlow, the structure of the network can be defined, the loss function (such as the mean squared error loss function) and the optimizer (such as the Adam optimizer) can be set, and then multiple iterations of training can be performed. For the test process, input the traffic light scheduling requirement test set into the trained network and calculate metrics such as the accuracy between the prediction result and the actual result. If the accuracy reaches a certain requirement (such as above 80%), it is considered that the model performance is good and it can be used as the traffic light scheduling requirement parsing model.

[0070] Through the training of the fully connected neural network, a traffic light scheduling requirement parsing model is obtained. This model can automatically predict the optimal traffic light scheduling requirement based on real-time traffic data, thereby realizing intelligent scheduling and providing accurate decision-making support for subsequent traffic light scheduling.

[0071] Furthermore, step S5 of the embodiment of the present application includes:

[0072] Step S51: Evaluate the matching degree between the traffic light scheduling requirement and the real-time traffic light scheduling plan to obtain a traffic light scheduling matching coefficient.

[0073] Step S52: Determine whether the traffic light scheduling matching coefficient meets the traffic light scheduling matching constraint.

[0074] Step S53: If the traffic light scheduling matching coefficient does not meet the traffic light scheduling matching constraint, generate a traffic light scheduling instruction.

[0075] Step S54: Based on the traffic light scheduling instruction, taking the traffic light scheduling requirement as the traffic light scheduling target, adjust the real-time traffic light scheduling plan to obtain the traffic light scheduling adjustment plan.

[0076] Specifically, perform a matching degree evaluation based on the real-time traffic light scheduling plan and the traffic light scheduling requirement obtained through the parsing model. By comparing the differences between the two in dimensions such as green light time, red light time, and traffic flow, calculate the traffic light scheduling matching coefficient, which is used to represent the matching degree between the current real-time traffic light scheduling plan and the predicted scheduling requirement. Usually, the matching coefficient is a value between 0 and 1, and the higher the value, the better the matching degree. Multiple methods can be used for the matching degree evaluation. For example, an evaluation index system can be established to quantitatively compare the key factors in the traffic light scheduling requirement and the real-time plan. For the factor of green light duration, if the recommended green light duration for a certain direction in the traffic light scheduling requirement is t1, and the green light duration in the real-time plan is t2, the absolute value of the difference between the two, ∣t1 - t2∣, can be calculated and incorporated into the overall matching degree calculation according to a certain weight. For the consideration of traffic flow in different directions, the ratio difference between the actual number of passing vehicles and the expected number of passing vehicles in the requirement can be calculated, etc. Combine various factors to obtain the traffic light scheduling matching coefficient.

[0077] The traffic light scheduling matching constraint is a preset standard or threshold used to determine whether the traffic light scheduling matching coefficient meets the requirements. Compare the traffic light scheduling matching coefficient with the traffic light scheduling matching constraint. If the matching coefficient is within this constraint range, it is considered that the current real-time traffic light scheduling plan basically meets the requirements; if not, adjustment is required.

[0078] When the traffic light scheduling matching coefficient does not meet the matching constraint, generate a traffic light scheduling instruction based on the traffic light scheduling requirement and the current matching situation, which is used to adjust the real-time traffic light scheduling plan. This instruction includes adjustment requirements for aspects such as green light duration, red light duration, and traffic light switching sequence. For example, if the traffic flow in a certain direction in the traffic light scheduling requirement is large, but the green light duration in the current real-time plan for this direction is short, then the traffic light scheduling instruction is to increase the green light duration for this direction. Decision trees or rule-based systems (conditional judgment and predefined adjustment rules) can be used to automatically generate these instructions. The generated traffic light scheduling instruction provides a clear operation basis for subsequent adjustment of the real-time traffic light scheduling plan, ensuring that the adjusted plan can better meet traffic demands.

[0079] Adjust the parameters in the real-time traffic light scheduling plan according to the traffic light scheduling instructions to obtain a traffic light scheduling adjustment plan. This scheduling adjustment plan can better adapt to the actual traffic conditions, improve the road traffic efficiency, and reduce traffic congestion. For example, if the traffic light scheduling instruction is to increase the green light duration in a certain direction, then find the corresponding green light duration parameter in the real-time traffic light scheduling plan and increase it according to the instruction. The adjustment of the plan can be implemented using the interface of the traffic signal control device or specialized traffic management software.

[0080] The above steps calculate the matching coefficient, determine whether the scheduling matching constraint is satisfied, and generate traffic light scheduling instructions for plan adjustment when it is not satisfied. The adjusted traffic light scheduling plan can more accurately meet the real-time traffic flow and demand, thereby reducing traffic congestion and improving the road traffic capacity.

[0081] In summary, the vehicle-road collaborative artificial intelligence traffic light optimization scheduling method provided by the embodiments of the present application has the following technical effects:

[0082] The embodiments of the present application first connect to the vehicle-road collaborative system to collect traffic data in real time and extract key spatio-temporal and dynamic feature information; then, through the analysis of these features, the current traffic light scheduling requirements are identified in real time. Next, the original traffic light scheduling plan is adjusted according to the identified real-time scheduling requirements to optimize the signal timing. Finally, by executing the optimized scheduling plan, intelligent traffic light scheduling based on dynamic traffic flow and road conditions changes is realized. Generally speaking, the embodiments of the present application can dynamically adjust the traffic light scheduling according to the actual traffic conditions, optimize the traffic resource allocation, thereby effectively improving the road traffic efficiency, reducing traffic congestion, and enhancing the utilization rate of road resources and the ability to cope with sudden traffic conditions.

[0083] Embodiment 2, as Figure 2 shown, based on the same inventive concept as the foregoing Embodiment 1, the embodiments of the present application provide a vehicle-road collaborative artificial intelligence traffic light optimization scheduling system, and the system includes:

[0084] A traffic data acquisition module 10, configured to connect to the vehicle-road collaborative system to obtain the road traffic data of the target road.

[0085] A feature sorting module 20, configured to sort features according to the road traffic data to determine the road traffic feature information.

[0086] A scheduling requirement identification module 30, configured to identify traffic light scheduling requirements according to the road traffic feature information to obtain traffic light scheduling requirements.

[0087] A plan acquisition module 40, configured to obtain the real-time traffic light scheduling plan of the target road.

[0088] The scheduling adjustment module 50 is used to adjust the real-time traffic light scheduling plan according to the traffic light scheduling requirements, and obtain a traffic light scheduling adjustment plan.

[0089] The scheduling execution module 60 is used to execute the optimized traffic light scheduling of the target road according to the traffic light scheduling adjustment plan.

[0090] Furthermore, the feature sorting module 20 of the embodiment of the present application is further used to execute the following steps:

[0091] Preprocess the road traffic data to obtain standard road traffic data; identify time features according to the standard road traffic data to obtain the first road traffic feature data; identify spatial features according to the standard road traffic data to obtain the second road traffic feature data; identify dynamic features according to the standard road traffic data to obtain the third road traffic feature data; perform data fusion according to the first road traffic feature data, the second road traffic feature data and the third road traffic feature data, and output the road traffic feature information.

[0092] Furthermore, the feature sorting module 20 of the embodiment of the present application is further used to execute the following steps:

[0093] Perform data cleaning on the road traffic data according to a predetermined cleaning factor to obtain the first processing result of traffic data; perform Kalman filtering on the first processing result of traffic data to obtain the second processing result of traffic data; perform normalization processing on the second processing result of traffic data to generate the standard road traffic data.

[0094] Furthermore, the predetermined cleaning factor includes outlier correction, duplicate value deletion and missing value filling.

[0095] Furthermore, the scheduling requirement identification module 30 of the embodiment of the present application is further used to execute the following steps:

[0096] Obtain a road traffic feature sample set and a traffic light scheduling requirement sample set; perform supervised learning according to the road traffic feature sample set and the traffic light scheduling requirement sample set to obtain a traffic light scheduling requirement parsing model; input the road traffic feature information into the traffic light scheduling requirement parsing model, and output the traffic light scheduling requirements.

[0097] Furthermore, the scheduling requirement identification module 30 of the embodiment of the present application is further used to execute the following steps:

[0098] Perform data cleaning on the road traffic feature sample set and the traffic light scheduling requirement sample set to obtain a traffic light scheduling requirement record set; perform data partitioning on the traffic light scheduling requirement record set to obtain a traffic light scheduling requirement training set and a traffic light scheduling requirement test set; train and test a fully connected neural network based on the traffic light scheduling requirement training set and the traffic light scheduling requirement test set to generate the traffic light scheduling requirement parsing model.

[0099] Further, the scheduling adjustment module 50 in the embodiment of the present application is further configured to perform the following steps:

[0100] Evaluate the matching degree between the traffic light scheduling requirement and the real-time traffic light scheduling plan to obtain a traffic light scheduling matching coefficient; determine whether the traffic light scheduling matching coefficient meets the traffic light scheduling matching constraint; if the traffic light scheduling matching coefficient does not meet the traffic light scheduling matching constraint, generate a traffic light scheduling instruction; based on the traffic light scheduling instruction, use the traffic light scheduling requirement as the traffic light scheduling target to adjust the real-time traffic light scheduling plan to obtain the traffic light scheduling adjustment plan.

[0101] Through the foregoing detailed description of the artificial intelligence traffic light optimization scheduling method for vehicle-road collaboration in this specification, those skilled in the art can clearly know the artificial intelligence traffic light optimization scheduling system for vehicle-road collaboration in this embodiment. For the system disclosed in Embodiment 2, since it corresponds to the method disclosed in Embodiment 1, it has corresponding functional modules and beneficial effects. For the relevant parts, refer to the description in the method section.

[0102] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. The vehicle-road collaborative artificial intelligence traffic light optimization scheduling method is characterized by: The method comprises: Connect to the vehicle-road cooperative system to obtain road traffic data of the target road; Perform feature sorting according to the road traffic data to determine road traffic feature information; Identify traffic light dispatching requirements based on the road traffic characteristic information to obtain traffic light dispatching requirements; Obtaining a real-time traffic light scheduling plan for the target road; Adjust the real-time traffic light scheduling plan according to the traffic light scheduling demand to obtain a traffic light scheduling adjustment plan; The traffic light optimization scheduling of the target road is performed according to the traffic light scheduling adjustment plan.

2. The vehicle-road collaborative artificial intelligence traffic light optimization scheduling method according to claim 1, characterized in that: The road traffic characteristic information is determined by performing feature sorting according to the road traffic data, including: Preprocessing the road traffic data to obtain road traffic standard data; Performing time feature recognition according to the road traffic standard data to obtain first road traffic feature data; Perform spatial feature recognition according to the road traffic standard data to obtain second road traffic feature data; Perform dynamic feature recognition according to the road traffic standard data to obtain third road traffic feature data; Data fusion is performed based on the first road traffic characteristic data, the second road traffic characteristic data and the third road traffic characteristic data, and the road traffic characteristic information is output.

3. The vehicle-road collaborative artificial intelligence traffic light optimization scheduling method according to claim 2, characterized in that: Preprocessing the road traffic data to obtain road traffic standard data includes: Performing data cleaning on the road traffic data according to a predetermined cleaning factor to obtain a first traffic data processing result; Performing Kalman filtering according to the first processing result of the traffic data to obtain a second processing result of the traffic data; Normalization processing is performed according to the second processing result of the traffic data to generate the road traffic standard data.

4. The vehicle-road collaborative artificial intelligence traffic light optimization scheduling method according to claim 3, characterized in that: The predetermined cleaning factors include outlier correction, duplicate value deletion and missing value filling.

5. The vehicle-road collaborative artificial intelligence traffic light optimization scheduling method according to claim 1, characterized in that: The traffic light dispatching requirement is identified according to the road traffic characteristic information to obtain the traffic light dispatching requirement, including: Obtaining a road traffic feature sample set and a traffic light scheduling demand sample set; Perform supervised learning based on the road traffic feature sample set and the traffic light scheduling demand sample set to obtain a traffic light scheduling demand parsing model; The road traffic characteristic information is input into the traffic light scheduling demand analysis model, and the traffic light scheduling demand is output.

6. The vehicle-road collaborative artificial intelligence traffic light optimization scheduling method according to claim 5, characterized in that: Performing supervised learning based on the road traffic feature sample set and the traffic light dispatch demand sample set to obtain a traffic light dispatch demand parsing model, including: Perform data cleaning according to the road traffic feature sample set and the traffic light dispatch demand sample set to obtain a traffic light dispatch demand record set; Data is divided according to the traffic light dispatching demand record set to obtain a traffic light dispatching demand training set and a traffic light dispatching demand test set; The fully connected neural network is trained and tested according to the traffic light scheduling requirement training set and the traffic light scheduling requirement test set to generate the traffic light scheduling requirement parsing model.

7. The vehicle-road collaborative artificial intelligence traffic light optimization scheduling method according to claim 1, characterized in that: The real-time traffic light scheduling scheme is adjusted according to the traffic light scheduling requirement to obtain a traffic light scheduling adjustment scheme, including: Performing a matching evaluation on the traffic light scheduling requirement and the traffic light scheduling real-time solution to obtain a traffic light scheduling matching coefficient; Determining whether the traffic light scheduling matching coefficient satisfies the traffic light scheduling matching constraint; If the traffic light scheduling matching coefficient does not satisfy the traffic light scheduling matching constraint, generating a traffic light scheduling instruction; Based on the traffic light scheduling instruction, the traffic light scheduling demand is used as the traffic light scheduling target, and the traffic light scheduling real-time plan is adjusted to obtain the traffic light scheduling adjustment plan.

8. The vehicle-road collaborative artificial intelligence traffic light optimization and dispatching system is characterized by: The system is used to execute the vehicle-road collaborative artificial intelligence traffic light optimization scheduling method according to any one of claims 1 to 7, comprising: Traffic data acquisition module, used to connect to the vehicle-road cooperative system and obtain road traffic data of the target road; A feature combing module, used to perform feature combing based on the road traffic data to determine road traffic feature information; A dispatch demand identification module, used to identify the traffic light dispatch demand according to the road traffic characteristic information, and obtain the traffic light dispatch demand; A scheme acquisition module, used to obtain a real-time traffic light scheduling scheme for the target road; A scheduling adjustment module, used to adjust the real-time traffic light scheduling plan according to the traffic light scheduling requirements to obtain a traffic light scheduling adjustment plan; A scheduling execution module is used to execute the traffic light optimization scheduling of the target road according to the traffic light scheduling adjustment plan.

Images