Intersection signal control method based on real-time traffic flow induction

Abstract

The invention discloses an intersection signal control method based on real-time traffic flow induction, and relates to the technical field of signal control, and the signal control method comprises the following steps: collecting congestion time, congestion length and congestion reasons of a current previous road section in different time periods; according to the intersection signal control method based on the real-time traffic flow induction, abnormal and missing values are removed through the signal control method, the data collection and preprocessing stage and interpolation processing, data completeness and accuracy are guaranteed, a foundation is built for follow-up work, during construction and training of a congestion prediction model, an LSTM layer effectively captures a time dependency relationship, and the traffic flow induction is effectively predicted. Through reasonable training and evaluation, it is ensured that the model accurately predicts the congestion condition, the traffic light time is dynamically adjusted, a timing scheme is individualized, a genetic algorithm is used for continuous optimization, the green light time can be flexibly and accurately allocated according to the real-time traffic flow, the intersection passing efficiency is greatly improved, and traffic congestion is effectively relieved.

Description

Technical Field

[0001] This invention relates to the field of signal control technology, specifically to a method for intersection signal control based on real-time traffic flow sensing. Background Technology

[0002] With the acceleration of urbanization, urban traffic flow is showing a rapid growth and increasingly complex and variable trend. Traditional intersection signal control methods are mostly based on fixed timing schemes, which typically use historical traffic flow data to set fixed traffic light durations within specific time periods.

[0003] In real-world traffic scenarios, traffic flow is affected by various factors, such as sudden traffic accidents, severe weather, and large-scale events. These factors can cause drastic fluctuations in traffic flow within a short period. Fixed traffic light timing schemes cannot adjust traffic light times in a timely manner to reflect these real-time changes, which can easily lead to increased traffic congestion on some road sections, while other sections may experience wasted green light time and excessively long waiting times, thus reducing overall traffic efficiency. To address this, we propose an intersection signal control method based on real-time traffic flow sensing. Summary of the Invention

[0004] The purpose of this invention is to provide an intersection signal control method based on real-time traffic flow sensing to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a cross-intersection signal control method based on real-time traffic flow sensing, the signal control method comprising the following steps: S1: Collect the congestion time, congestion length, and congestion cause of the current road segment at different time periods, remove outliers and missing values, integrate historical congestion data and real-time traffic data to form a unified dataset, extract key features from the dataset as model input; S2: Build a congestion prediction model using an LSTM model, train it using historical data, and evaluate the performance of the congestion prediction model through cross-validation; S3: Based on the output of the congestion prediction model and combined with the current real-time traffic data, use a genetic algorithm to dynamically adjust the traffic light timings.

[0006] Preferably, the key features in step S1 are: congestion time, congestion length, traffic flow, and vehicle speed.

[0007] Preferably, the congestion prediction model construction in step S2 includes the following steps: S2.1: Define the input dimension of the model as the number of extracted features, and add an LSTM layer to capture the time dependencies in the sequence data; S2.2: Define the output dimension of the model as the predicted congestion time or congestion length, and use a linear activation function to output continuous values. S2.3: Select the mean squared error (MSE) to train the congestion prediction model. Divide the integrated dataset into a training set, a validation set, and a test set. Use the training set to iteratively train the model. After each iteration, use the validation set to evaluate the performance of the congestion prediction model.

[0008] Preferably, in step S2.1, the congestion prediction model can be optimized by adjusting the number of neurons or the activation function of the LSTM layer.

[0009] Preferably, in step S2.3, the trained model is evaluated using a test set. The prediction accuracy of the model can be evaluated by calculating the error index between the predicted value and the true value.

[0010] Preferably, step S3 specifically includes: treating the traffic light time allocation scheme as an individual, discretizing the green light time length for each direction, and then representing it with binary numbers; for multi-phase intersections, combining the green light times for each phase to form a complete individual; evaluating the fitness value of each individual according to the tournament selection method function, selecting excellent individuals to enter the next generation, performing crossover operations on the selected individuals to generate new individuals; the crossover operation can randomly select a portion of the genes of two individuals to exchange to generate new gene combinations; performing mutation operations on the crossover individuals to increase the diversity of the population; continuously iterating and executing selection, crossover, and mutation operations until the termination condition is met, and applying the final optimized solution to actual traffic signal control.

[0011] Preferably, the termination condition for continuous iteration is: reaching the target fitness value.

[0012] Preferably, in step S1, outliers and missing values ​​in the data are removed by interpolation.

[0013] Preferably, in step S1, historical congestion data and real-time traffic data are time-aligned and spatially matched to form a unified dataset, ensuring that the temporal granularity and spatial range of the data are consistent.

[0014] Preferably, in step S3, in each iteration, the fitness value of each individual is evaluated using a fitness function, and genetic operations are performed based on the evaluation results.

[0015] Compared with the prior art, the beneficial effects of the present invention are: This invention utilizes a signal control method. In the data collection and preprocessing stage, interpolation removes outliers and missing values, ensuring data integrity and accuracy, thus laying a solid foundation for subsequent work. During the construction and training of the congestion prediction model, the LSTM layer effectively captures time dependencies, and reasonable training and evaluation ensure that the model accurately predicts congestion. In the dynamic adjustment stage of traffic light timing, the timing scheme is individualized and continuously optimized using a genetic algorithm, enabling flexible and accurate allocation of green light time based on real-time traffic flow, greatly improving the efficiency of intersection passage and effectively alleviating traffic congestion. Attached Figure Description

[0016] Figure 1 This is a flowchart of the signal control method of the present invention. Detailed Implementation

[0017] 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. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Example: Please refer to Figure 1 This invention provides a technical solution: a cross-intersection signal control method based on real-time traffic flow sensing, the signal control method comprising the following steps: S1: Collect the congestion time, congestion length, and congestion cause of the current road segment at different time periods, remove outliers and missing values, integrate historical congestion data and real-time traffic data to form a unified dataset, extract key features from the dataset as model input; S2: Build a congestion prediction model using an LSTM model, train it using historical data, and evaluate the performance of the congestion prediction model through cross-validation; S3: Based on the output of the congestion prediction model and combined with the current real-time traffic data, use a genetic algorithm to dynamically adjust the traffic light timings.

[0019] The key features in step S1 are: congestion time, congestion length, traffic flow, and vehicle speed.

[0020] The congestion prediction model construction in step S2 includes the following steps: S2.1: Define the input dimension of the model as the number of extracted features, and add an LSTM layer to capture the time dependencies in the sequence data; S2.2: Define the output dimension of the model as the predicted congestion time or congestion length, and use a linear activation function to output continuous values. S2.3: Select the mean squared error (MSE) to train the congestion prediction model. Divide the integrated dataset into a training set, a validation set, and a test set. Use the training set to iteratively train the model. After each iteration, use the validation set to evaluate the performance of the congestion prediction model.

[0021] In step S2.1, the congestion prediction model can be optimized by adjusting the number of neurons or the activation function of the LSTM layer.

[0022] In step S2.3, the trained model is evaluated using a test set. The prediction accuracy of the model can be assessed by calculating the error metric between the predicted and true values.

[0023] Step S3 specifically includes: treating the traffic light time allocation scheme as an individual, discretizing the green light duration for each direction and representing it with binary numbers; for multi-phase intersections, combining the green light durations for each phase to form a complete individual; evaluating the fitness value of each individual according to the tournament selection method function, selecting excellent individuals to enter the next generation, performing crossover operations on the selected individuals to generate new individuals; the crossover operation can randomly select a portion of the genes of two individuals to exchange to generate new gene combinations; performing mutation operations on the crossover individuals to increase the diversity of the population; continuously iterating the selection, crossover, and mutation operations until the termination condition is met; and finally applying the optimized solution to actual traffic signal control.

[0024] The termination condition for continuous iteration is reaching the target fitness value.

[0025] In step S1, outliers and missing values ​​are removed from the data through interpolation.

[0026] In step S1, historical congestion data and real-time traffic data are time-aligned and spatially matched to form a unified dataset, ensuring that the temporal granularity and spatial range of the data are consistent.

[0027] In step S3, in each iteration, the fitness function is used to evaluate the fitness value of each individual, and genetic operations are performed based on the evaluation results.

[0028] The specific implementation method of this embodiment is as follows: In the data collection and preprocessing stage, data on congestion time, length, and causes at different time periods for road segments are first collected. Outliers and missing values ​​are removed through interpolation. Then, historical congestion data and real-time traffic data are aligned temporally and matched spatially to form a unified dataset, ensuring consistency in temporal granularity and spatial range. Congestion time, length, traffic flow, and vehicle speed are extracted as key features input to the model. In the congestion prediction model construction and training stage, the model input dimension is defined as the number of extracted features. An LSTM layer is added to capture temporal dependencies, and optimization can be achieved by adjusting the number of neurons or the activation function. The output dimension is defined as the predicted congestion time or length, and a linear activation function is used to output continuous values. Mean Squared Error (MSE) is selected to train the model. The integrated dataset is divided into training, validation, and test sets, and iterative training is performed using the training set. After each iteration, the model is evaluated using a validation set, and finally, the model's prediction accuracy is evaluated using a test set by calculating the error index between the predicted and actual values. During the dynamic adjustment phase of traffic light times, the traffic light time allocation scheme is treated as an individual. The green light time length for each direction is discretized and represented by binary numbers. The green light times of each phase at multi-phase intersections are combined to form a complete individual. The fitness value of the individual is evaluated according to the tournament selection method function, and the best individuals are selected to enter the next generation. The selected individuals are subjected to crossover (randomly selecting two individuals to exchange some genes to generate new gene combinations) and mutation (increasing population diversity) operations. In each iteration, the fitness value of the individual is evaluated using a fitness function and genetic operations are performed accordingly. The selection, crossover, and mutation operations are continuously executed iteratively until the target fitness value is reached, which is the termination condition. The final optimized solution is then applied to actual traffic signal control.

[0029] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0030] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for intersection signal control based on real-time traffic flow sensing, characterized in that: The signal control method includes the following steps: S1: Collect the congestion time, congestion length, and congestion cause of the current road segment at different time periods, remove outliers and missing values, integrate historical congestion data and real-time traffic data to form a unified dataset, extract key features from the dataset as model input; S2: Build a congestion prediction model using an LSTM model, train it using historical data, and evaluate the performance of the congestion prediction model through cross-validation; S3: Based on the output of the congestion prediction model and combined with the current real-time traffic data, use a genetic algorithm to dynamically adjust the traffic light timings.

2. The intersection signal control method based on real-time traffic flow sensing according to claim 1, characterized in that: The key features in step S1 are: congestion time, congestion length, traffic flow, and vehicle speed.

3. The intersection signal control method based on real-time traffic flow sensing according to claim 1, characterized in that: The congestion prediction model construction in step S2 includes the following steps: S2.1: Define the input dimension of the model as the number of extracted features, and add an LSTM layer to capture the time dependencies in the sequence data; S2.2: Define the output dimension of the model as the predicted congestion time or congestion length, and use a linear activation function to output continuous values. S2.3: Select the mean squared error (MSE) to train the congestion prediction model. Divide the integrated dataset into a training set, a validation set, and a test set. Use the training set to iteratively train the model. After each iteration, use the validation set to evaluate the performance of the congestion prediction model.

4. The intersection signal control method based on real-time traffic flow sensing according to claim 3, characterized in that: In step S2.1, the congestion prediction model can be optimized by adjusting the number of neurons or the activation function of the LSTM layer.

5. The intersection signal control method based on real-time traffic flow sensing according to claim 3, characterized in that: In step S2.3, the trained model is evaluated using a test set. The prediction accuracy of the model can be evaluated by calculating the error index between the predicted value and the true value.

6. The intersection signal control method based on real-time traffic flow sensing according to claim 1, characterized in that: Step S3 specifically includes: treating the traffic light time allocation scheme as an individual, discretizing the green light time length for each direction, and then representing it with binary numbers. For multi-phase intersections, the green light times for each phase are combined to form a complete individual. The fitness value of each individual is evaluated according to the tournament selection method function, and excellent individuals are selected to enter the next generation. Crossover operations are performed on the selected individuals to generate new individuals. The crossover operation can randomly select a portion of the genes of two individuals to exchange to generate new gene combinations. Mutation operations are performed on the crossover individuals to increase the diversity of the population. The selection, crossover, and mutation operations are continuously iterated until the termination condition is met, and the final optimized solution is applied to actual traffic signal control.

7. The intersection signal control method based on real-time traffic flow sensing according to claim 6, characterized in that: The termination condition for continuous iteration is reaching the target fitness value.

8. The intersection signal control method based on real-time traffic flow sensing according to claim 1, characterized in that: In step S1, outliers and missing values ​​in the data are removed by interpolation.

9. The intersection signal control method based on real-time traffic flow sensing according to claim 1, characterized in that: In step S1, historical congestion data and real-time traffic data are time-aligned and spatially matched to form a unified dataset, ensuring that the temporal granularity and spatial range of the data are consistent.

10. A method for intersection signal control based on real-time traffic flow sensing according to claim 6, characterized in that: In each iteration of step S3, the fitness function is used to evaluate the fitness value of each individual, and genetic operations are performed based on the evaluation results.

Images