Multi-intersection signal lamp cooperative control method based on stepwise regression and Webster

Abstract

The invention provides a multi-intersection signal lamp cooperative control method based on stepwise regression and Webster. The multi-intersection signal lamp cooperative control method comprises the following steps: acquiring initial traffic flow data of each traffic intersection and preprocessing the initial traffic flow data; defining physical vectors of different motion modes of vehicles, using a vector statistical method to count the traffic flow of different physical vectors of each traffic intersection, and constructing a multivariate vector stepwise regression model to predict the total traffic flow of each traffic intersection; based on the total traffic flow of each traffic intersection in unit time, optimizing the traffic light duration of the signal lamps by using a Webster timing method, and performing cooperative control on the signal lamps at multiple intersections according to an optimization result; according to the method, on the basis of unknown traffic preposition information and a classic machine learning mode, the traffic flow of the whole intersection of the main road is predicted through the vector stepwise regression model, then the signal lamps are adaptively controlled and optimized in combination with the Webster timing method, and the method has the advantages of high benefit, high credibility and high robustness.

Description

Technical Field

[0001] The present invention relates to the technical field of traffic light control, and more specifically, to a multi-intersection signal lamp collaborative control method based on stepwise regression and Webster. Background Art

[0002] Adaptive control of signal lamps is an important topic in the traffic field and a key link in the construction of smart cities. With the development of deep learning, research on optimizing signal lamp control based on neural networks has gradually emerged. Researchers have built large neural networks to simulate the non-linear dynamic characteristics of traffic systems.

[0003] Most traditional studies only focus on single main road intersections. However, the real complex traffic does not only revolve around a single intersection. At present, although there are some existing technologies that have optimized the control of multi-intersection signal lamps, there are still defects. For example, the patent document with the publication number "CN109697867A" provides a traffic control method and system based on deep learning. The method includes: a local processor obtains video data of a traffic intersection, analyzes and calculates traffic data using a deep neural network, and each local processor in each mode selects to control the signal lamp by itself or optimize the signal lamp setting in combination with a city central server according to the traffic data. However, most existing technologies including this solution still require prior information such as known signal cycles, red and green light timings, vehicle queues, and green signal ratios. The acquisition and setting of traffic prior information rely on a large amount of empirical data and are difficult to effectively optimize under unknown conditions. In addition, the sharp increase in data volume caused by traffic saturation and the high computational requirements of deep neural networks are likely to cause out-of-memory (OOM) problems, significantly increasing the computational resource cost.

[0004] Therefore, it is of great theoretical and practical significance to study how to use machine learning methods to achieve adaptive collaborative control of signal lamps at multiple intersections on the main road under the condition of unknown prior traffic information. Summary of the Invention

[0005] In order to overcome the defects that the above-mentioned existing technologies are difficult to effectively optimize under unknown conditions and the computational cost of deep learning models is high, the present invention provides a multi-intersection signal lamp collaborative control method based on stepwise regression and Webster. Based on the classical machine learning mode on the basis of unknown traffic prior information, first, a vector stepwise regression model is used to predict the traffic flow of all intersections on the main road, and then an adaptive timing model and Webster timing method are combined to perform adaptive control simulation optimization on the signal lamps, taking into account both model benefits and robustness, and being able to effectively optimize the traffic flow of all intersections on the main road in multiple time periods.

[0006] To solve the above technical problems, the technical solution of the present invention is as follows: A multi-intersection signal light coordinated control method based on stepwise regression and Webster includes the following steps: S1: Collect the initial traffic flow data of each traffic intersection and perform preprocessing; S2: Define the physical vectors of different vehicle motion modes, use vector statistics to count the traffic flow of different physical vectors at each traffic intersection, and build a multivariate vector stepwise regression model to predict the overall traffic flow at each traffic intersection; S3: Based on the overall traffic flow per unit time at each traffic intersection, the Webster timing method is used to optimize the traffic light duration of each traffic intersection, and the traffic lights at multiple intersections are coordinated and controlled according to the optimization results.

[0007] Preferably, in step S1, cameras at various traffic intersections are used to collect initial traffic flow data in various directions of the corresponding intersections, and the initial traffic flow data at least includes: traffic flow image data, vehicle travel direction, shooting time, shooting location and license plate number of each vehicle.

[0008] Preferably, the preprocessing includes: using the Pandas library to perform outlier detection on the initial traffic flow data and remove abnormal data.

[0009] Preferably, in step S2, the movement mode of the vehicle is the relative displacement direction of the vehicle from one traffic intersection to the next traffic intersection, including straight driving, left turn, right turn and U-turn; Four different physical vectors are used to represent the vehicle's four motion modes: straight ahead, left turn, right turn, and U-turn, recorded as , , and .

[0010] Preferably, based on the preprocessed initial traffic flow data, the vehicle's movement mode is determined by tracking the physical vector directions of the vehicle at the current traffic intersection and the next traffic intersection, and the traffic flow corresponding to each physical vector at the current traffic intersection is counted.

[0011] Preferably, in step S2, the multivariate vector stepwise regression model constructed is specifically:

[0012] in, The prediction result of the overall traffic flow at the traffic intersection; , , and are the first, second, third and fourth weights respectively; is the preset parameter; Train the multivariate vector stepwise regression model based on the least squares method, and use the trained multivariate vector stepwise regression model to predict the overall traffic flow at each traffic intersection.

[0013] Preferably, in step S3, calculate the overall traffic flow per unit time at each traffic intersection according to the following formula :

[0014] where is the time the overall traffic flow passing through the traffic intersection within; is the unit time.

[0015] Preferably, in step S3, use the Webster timing method to optimize the red and green light durations of the traffic lights at each traffic intersection, including: Data exploratory analysis: Based on the preprocessed initial traffic flow data, count the daily traffic flow at each traffic intersection and perform secondary mean processing, and visualize the daily traffic flow at each traffic intersection and the results of its secondary mean processing respectively; Inference of red and green light durations: Based on the results of data exploratory analysis, infer the initial red light duration and initial green light duration of the traffic light; Use the Webster timing method to optimize and adjust the green light duration of the traffic lights at each traffic intersection, obtain the optimized green light duration, and control the traffic lights according to the optimized green light duration.

[0016] Preferably, the inference of the red and green light durations includes: When the stop time interval of the i-th vehicle is greater than the preset threshold , it is determined that the vehicle encounters a red light, and count the initial red light duration:

[0017] where is the initial red light duration; According to the initial red light duration calculate the initial green light duration:

[0018] where is the initial green light duration; is the signal cycle.

[0019] Preferably, calculate the optimized green light duration according to the following formula:

[0020] where is the optimized green light duration, is the initial green light duration occupies the signal cycle proportion; According to the optimized green light duration control the signal lights, and calculate the overall traffic flow per unit time at each traffic intersection after optimization according to the following formula :

[0021] According to the overall traffic flow per unit time after optimization evaluate the optimization effect of the traffic efficiency at each traffic intersection.

[0022] Compared with the prior art, the beneficial effects of the technical solution of the present invention are: The present invention provides a multi-intersection signal light cooperative control method based on stepwise regression and Webster. First, collect the initial traffic flow data of each traffic intersection and perform preprocessing; then define the physical vectors of different vehicle movement modes, use the vector statistical method to count the traffic flow of different physical vectors at each traffic intersection, and construct a multiple vector stepwise regression model to predict the overall traffic flow at each traffic intersection; finally, based on the overall traffic flow per unit time at each traffic intersection, use the Webster timing method to optimize the red and green light durations of the signal lights at each traffic intersection, and perform cooperative control on the multi-intersection signal lights according to the optimization results; The present invention can perform data mining on the vehicle state information at traffic intersections based on unknown pre-traffic information, accurately distinguish the vector phases of vehicles at traffic intersections, and perform adaptive control optimization on the traffic flow at multiple intersections of the main road by independently inferring the red and green light durations, thereby improving traffic efficiency; secondly, the present invention is based on a data-driven mode, is applicable to the complex traffic conditions of all intersections of the main road, has good robustness, and can be directly plugged and used; in addition, the present invention is based on a classic machine learning mode. Compared with deep learning, the present invention saves a large amount of computing resources and has the advantage of high efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 is a flow chart of a multi-intersection signal light cooperative control method based on stepwise regression and Webster provided in Embodiment 1.

[0024] Figure 2 is a framework diagram of a multi-intersection signal light cooperative control method based on stepwise regression and Webster provided in Embodiment 2.

[0025] Figure 3 is a schematic diagram of physical vectors provided in Embodiment 2. DETAILED DESCRIPTION OF THE INVENTION

[0026] The accompanying drawings are only for illustrative purposes and should not be construed as a limitation to this application; For better illustration of this embodiment, some components in the accompanying drawings are omitted, enlarged or reduced, which do not represent the dimensions of the actual product; For those skilled in the art, it is understandable that some well-known structures and their descriptions in the accompanying drawings may be omitted.

[0027] The technical solutions of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0028] Embodiment 1 As Figure 1 shown, this embodiment provides a multi-intersection signal lamp cooperative control method based on stepwise regression and Webster, including the following steps: S1: Collect the initial traffic flow data of each traffic intersection and perform preprocessing; S2: Define the physical vectors of different vehicle movement modes, use the vector statistical method to count the traffic flow of different physical vectors at each traffic intersection, and construct a multiple vector stepwise regression model to predict the overall traffic flow at each traffic intersection; S3: Based on the overall traffic flow at each traffic intersection per unit time, use the Webster timing method to optimize the red and green light durations of the signal lamps at each traffic intersection, and perform cooperative control on the multi-intersection signal lamps according to the optimization results.

[0029] In the specific implementation process, first collect the initial traffic flow data of each traffic intersection and perform preprocessing; Then define the physical vectors of different vehicle movement modes, use the vector statistical method to count the traffic flow of different physical vectors at each traffic intersection, and construct a multiple vector stepwise regression model to predict the overall traffic flow at each traffic intersection; Finally, based on the overall traffic flow at each traffic intersection per unit time, use the Webster timing method to optimize the red and green light durations of the signal lamps at each traffic intersection, and perform cooperative control on the multi-intersection signal lamps according to the optimization results; Based on the unknown traffic pre-information, this method is based on the classical machine learning mode. First, the traffic flow of all intersections on the main road is predicted through a vector stepwise regression model, and then combined with the adaptive timing model and the Webster timing method to perform adaptive control simulation optimization on the signal lamps, taking into account both the model benefits and robustness, and can effectively optimize the traffic flow of all intersections on the main road in multiple time periods.

[0030] Embodiment 2 This embodiment provides a multi-intersection signal lamp cooperative control method based on stepwise regression and Webster, including the following steps: S1: Collect the initial traffic flow data of each traffic intersection and perform preprocessing; S2: Define the physical vectors of different vehicle motion modes, use vector statistics to count the traffic flow of different physical vectors at each traffic intersection, and build a multivariate vector stepwise regression model to predict the overall traffic flow at each traffic intersection; S3: Based on the overall traffic flow per unit time at each traffic intersection, the Webster timing method is used to optimize the duration of traffic lights at each traffic intersection, and the traffic lights at multiple intersections are coordinated and controlled according to the optimization results; In the step S1, the cameras at each traffic intersection are used to collect initial traffic flow data in each direction of the corresponding intersection, and the initial traffic flow data at least includes: traffic flow image data, vehicle travel direction, shooting time, shooting location and license plate number of each vehicle; The preprocessing includes: using the Pandas library to perform outlier detection on the initial traffic flow data and remove abnormal data; In step S2, the movement mode of the vehicle is the relative displacement direction of the vehicle from one traffic intersection to the next traffic intersection, including straight driving, left turn, right turn and U-turn; Four different physical vectors are used to represent the vehicle's four motion modes: straight ahead, left turn, right turn, and U-turn, recorded as , , and ; Based on the pre-processed initial traffic flow data, the vehicle's movement mode is determined by tracking the physical vector direction of the vehicle at the current traffic intersection and the next traffic intersection, and the traffic flow corresponding to each physical vector at the current traffic intersection is counted; In step S2, the multivariate vector stepwise regression model constructed is specifically:

[0031] in, The prediction result of the overall traffic flow at the traffic intersection; , , and are the first, second, third and fourth weights respectively; is the preset parameter; Training the multivariate vector stepwise regression model based on the least squares method, and using the trained multivariate vector stepwise regression model to predict the overall traffic flow at each traffic intersection; In step S3, the total traffic flow per unit time at each traffic intersection is calculated according to the following formula: :

[0032] Among them, is the total traffic volume passing through the intersection within a certain period of time; is the unit time; In step S3, the Webster timing method is used to optimize the red and green light durations of the signal lights at each traffic intersection, including: Data exploratory analysis: Based on the preprocessed initial traffic flow data, the traffic volume of each traffic intersection is counted every day and secondary mean processing is performed, and the traffic volume of each traffic intersection every day and the results of its secondary mean processing are visualized respectively; Inference of red and green light durations: Based on the results of data exploratory analysis, the initial red light duration and initial green light duration of the signal lights are inferred; Use the Webster timing method to optimize and adjust the green light duration of the signal lights at each traffic intersection, obtain the optimized green light duration, and control the signal lights according to the optimized green light duration; The inference of the red and green light durations includes: When the stop time interval of the i-th vehicle is greater than the preset threshold , it is determined that the vehicle encounters a red light, and the initial red light duration is counted:

[0033] Among them, is the initial red light duration; According to the initial red light duration calculate the initial green light duration:

[0034] Among them, is the initial green light duration; is the signal light cycle; Calculate the optimized green light duration according to the following formula:

[0035] Among them, is the optimized green light duration, is the proportion of the initial green light duration in the signal light cycle ; Control the signal lights according to the optimized green light duration , and calculate the optimized total traffic volume per unit time of each traffic intersection according to the following formula :

[0036] According to the optimized total traffic volume per unit time Evaluate the optimization effect of the traffic efficiency at each traffic intersection.

[0037] In the specific implementation process, such as Figure 2 shown, first collect the initial traffic flow data of each traffic intersection and perform preprocessing; Specifically, first use the cameras at each traffic intersection to collect the initial traffic flow data in all directions of the corresponding intersection, including: traffic flow image data, vehicle driving direction, shooting time, shooting location, and license plate numbers of each vehicle; then use the Pandas library to detect outliers in the initial traffic flow data and remove the abnormal data to ensure the accuracy and reliability of the dataset; Meanwhile, conduct exploratory data analysis, count the daily traffic flow data of each intersection, and visualize it; considering that the data shows a certain degree of volatility, perform secondary mean processing and visualization to obtain the dynamic change process of the traffic flow, and thus provide corresponding decision-making assistance for dynamic control optimization to complete the preprocessing; Next, define the physical vectors of different vehicle motion modes, use the vector statistical method to count the traffic flow of different physical vectors at each traffic intersection, and construct a multiple vector stepwise regression model to predict the overall traffic flow at each traffic intersection; In this embodiment, "phase" is used to define the relative displacement direction taken by a vehicle when driving from one intersection to the next on the road; "vector", as an important concept in mathematics and physics, includes two basic attributes: magnitude and direction; in this embodiment, the directionality of the vector is fused with the phase of the vehicle, and this definition covers four basic motion modes to clarify the motion trajectory and direction selection of the vehicle at the intersection; such as Figure 3 shown, use four different physical vectors to represent the four motion modes of the vehicle going straight, turning left, turning right, and making a U-turn, denoted as , , and ; The vector statistical method is essentially a data analysis means. On the basis of determining the vector phase of the vehicle, by recording the driving direction of the vehicle at the starting intersection and tracking the vector direction where it is photographed at the next consecutive intersection, to judge and count the specific vector phase operations completed by the vehicle; when the pre-traffic information is unknown, obtain the vehicle data of different motion modes at each intersection through vector statistics; After determining the number of vehicles in different vector phases at each intersection on the main road, construct a vector stepwise regression model to accurately predict the overall traffic flow at each intersection, which plays a key role in setting the control cycle of the traffic lights; the constructed multiple vector stepwise regression model is specifically:

[0038] Among them, is the prediction result of the overall traffic flow at a traffic intersection; , , and are the first, second, third, and fourth weights respectively; is a preset parameter; Train the multivariate vector stepwise regression model based on the least squares method, and use the trained multivariate vector stepwise regression model to predict the overall traffic flow at each traffic intersection; Finally, based on the overall traffic flow at each traffic intersection per unit time, use the Webster timing method to optimize the red and green light durations of the traffic lights at each traffic intersection, and perform coordinated control on the multi-intersection traffic lights according to the optimization results; this step only uses the initial traffic flow information of the known intersections, infers the traffic flow and red and green light timings of the entire intersection based on the vector stepwise regression model, so as to achieve adaptive simulation optimization; Specifically, first, it is necessary to obtain the traffic flow at the intersection per unit time, and calculate the overall traffic flow per unit time by using the prediction result output by the vector stepwise regression model :

[0039] wherein, is the time within which the overall traffic flow passes through the traffic intersection; is the unit time (for example, 1 minute); After that, based on the data exploratory analysis results, infer the initial red light duration and the initial green light duration of the traffic lights; The inference of the red and green light durations needs to be calculated in combination with the vehicle status information data obtained by the traffic intersection camera. Specifically, when the stop time interval of the i-th vehicle is greater than the preset threshold , it is determined that the vehicle has encountered a red light, and the initial red light duration is statistically calculated. The formula is as follows:

[0040] wherein, is the initial red light duration; In this embodiment, the "stop time interval" is judged according to the time when the license plate number is photographed by each camera for its time interval at a single intersection. If it is photographed multiple times at a single intersection and the time interval before and after is greater than the preset threshold, it is considered that it has encountered a red light; Since the signal light cycle is fixed, the initial green light duration can be calculated according to the initial red light duration :

[0041] Among them, is the initial green light duration; is the signal light cycle; Finally, the Webster timing method is used to optimize and adjust the green light duration of the signal lights at each traffic intersection, obtain the optimized green light duration, and control the signal lights according to the optimized green light duration to further improve traffic efficiency; Webster is a classic traffic engineering tool. It determines the optimal signal light timing by considering factors such as the traffic flow at the intersection (the number of vehicles passing through the intersection per minute), the red and green light cycle time (usually a fixed value, such as 120 seconds), and the proportion of the green light time in the cycle time. The optimized green light duration is calculated according to the following formula:

[0042] Among them, is the optimized green light duration, is the initial green light duration occupies the signal light cycle proportion, in this embodiment, is initialized to 0.65 for parameter initialization of Webster simulation; According to the optimized green light duration control the signal lights, and calculate the overall traffic flow per unit time optimized for each traffic intersection according to the following formula :

[0043] According to the overall traffic flow per unit time optimized evaluate the optimization effect of the traffic efficiency at each traffic intersection; In this embodiment, it is necessary to continuously simulate and optimize in the simulation system, and evaluate the traffic efficiency in real time, and continuously adjust the parameters to obtain the best green light time setting; Through the above formula, the ideal green light time required for each intersection under different traffic flow conditions can also be calculated; in addition, the average speed of the traffic flow can be optimized by adjusting the green light time, thereby reducing traffic congestion and improving road use efficiency; through this method, not only can the traffic efficiency of a single intersection be improved, but also a better traffic flow distribution can be achieved in the entire traffic network, improving road use efficiency; Based on the unknown traffic pre-information and the classical machine learning mode, this method first predicts the traffic flow of all intersections on the main road through a vector stepwise regression model, and then combines an adaptive timing model and the Webster timing method to simulate and optimize the signal lights adaptively. It takes into account both the model benefits and robustness, and can effectively optimize the traffic flow of all intersections on the main road in multiple time periods.

[0044] Like or similar reference numerals correspond to like or similar components; The terms describing the positional relationship in the drawings are only for illustrative purposes and should not be construed as a limitation to this application; Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, rather than limiting the implementation manners of the present invention. For those of ordinary skill in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to enumerate all the implementation manners here. Any modifications, equivalent replacements, and improvements made within the spirit and principle of the present invention shall be included within the protection scope of the claims of the present invention.

Claims

1. A multi-intersection signal light collaborative control method based on stepwise regression and Webster, characterized in that The following steps are involved: S1: Collect the initial traffic flow data of each traffic intersection and perform preprocessing; S2: Define the physical vectors of different vehicle motion modes, use vector statistics to count the traffic flow of different physical vectors at each traffic intersection, and build a multivariate vector stepwise regression model to predict the overall traffic flow at each traffic intersection; S3: Based on the overall traffic flow per unit time at each traffic intersection, the Webster timing method is used to optimize the traffic light duration of each traffic intersection, and the traffic lights at multiple intersections are coordinated and controlled according to the optimization results.

2. The multi-intersection signal lamp collaborative control method based on stepwise regression and Webster according to claim 1, wherein, In step S1, the cameras at each traffic intersection are used to collect initial traffic flow data in each direction of the corresponding intersection, and the initial traffic flow data at least includes: traffic flow image data, vehicle travel direction, shooting time, shooting location and license plate number of each vehicle.

3. The multi-intersection signal lamp collaborative control method based on stepwise regression and Webster according to claim 1, characterized in that The preprocessing includes: using the Pandas library to perform outlier detection on the initial traffic flow data and remove abnormal data.

4. A multi-intersection signal lamp collaborative control method based on stepwise regression and Webster according to claim 1, characterized in that In step S2, the movement mode of the vehicle is the relative displacement direction of the vehicle from one traffic intersection to the next traffic intersection, including straight driving, left turn, right turn and U-turn; Four different physical vectors are used to represent the four motion modes of a vehicle, namely going straight, turning left, turning right, and doing a U-turn, denoted as , , and .

5. A multi-intersection signal lamp collaborative control method based on stepwise regression and Webster according to claim 4, characterized in that, Based on the preprocessed initial traffic flow data, the vehicle's movement mode is determined by tracking the physical vector directions of the vehicle at the current traffic intersection and the next traffic intersection, and the traffic flow corresponding to each physical vector at the current traffic intersection is counted.

6. The multi-intersection signal lamp collaborative control method based on stepwise regression and Webster according to claim 5, wherein In step S2, the multivariate vector stepwise regression model constructed is specifically: Among them, is the predicted result of the overall traffic flow at the traffic intersection; , , and are the first, second, third, and fourth weights respectively; is a preset parameter; The multivariate vector stepwise regression model is trained based on the least squares method, and the overall traffic flow at each traffic intersection is predicted using the trained multivariate vector stepwise regression model.

7. A multi-intersection signal lamp collaborative control method based on stepwise regression and Webster according to claim 6, characterized in that, In the step S3, the overall traffic flow per unit time of each traffic intersection is calculated according to the following formula : Among them, is the total traffic volume passing through the intersection within a period of time; is the unit time.

8. A multi-intersection signal lamp collaborative control method based on stepwise regression and Webster according to claim 7, characterized in that, In step S3, the Webster timing method is used to optimize the duration of traffic lights at each traffic intersection, including: Data exploratory analysis: Based on the pre-processed initial traffic flow data, the daily traffic flow of each traffic intersection is counted and processed by the secondary mean, and the daily traffic flow of each traffic intersection and its secondary mean processing results are visualized respectively; Traffic light duration inference: Based on the results of data exploratory analysis, the initial red light duration and initial green light duration of the traffic light are inferred; The Webster timing method is used to optimize and adjust the green light duration of traffic lights at each traffic intersection, obtain the optimized green light duration, and control the traffic lights according to the optimized green light duration.

9. A multi-intersection signal lamp collaborative control method based on stepwise regression and Webster according to claim 8, characterized in that The traffic light duration inference includes: When the stop time interval of the i-th vehicle is greater than a preset threshold it is determined that the vehicle has encountered a red light, and the initial red light duration is counted: Among them, is the initial red light duration; According to the initial red light duration Calculate the initial green light duration: Among them, is the initial green light duration; is the signal light cycle.

10. A multi-intersection signal lamp cooperative control method based on stepwise regression and Webster according to claim 9, characterized in that, The optimized green light duration is calculated according to the following formula: Among them, is the optimized green light duration, is the initial green light duration occupies the signal light cycle ratio; According to the optimized green light duration Control the signal lights and calculate the overall traffic flow per unit time at each intersection after optimization according to the following formula : According to the overall traffic flow per unit time after optimization Evaluate the optimization effect of the traffic efficiency at each traffic intersection.

Images