Urban intelligent traffic monitoring and early warning method and device based on artificial intelligence, and medium

Abstract

The invention relates to the technical field of intelligent early warning, and discloses an artificial intelligence urban intelligent traffic monitoring early warning method and device and a medium, and the method comprises the steps: obtaining a traffic monitoring image collected by any signal intersection of a city in real time; generating a traffic condition prediction model and evaluating the influence of the current road condition information and weather information on traffic; abnormal behaviors in the traffic flow are identified, and early warning is carried out in advance; executing a dynamic traffic signal lamp period adjusting strategy, and adjusting the period of the traffic signal lamp in real time; according to the real-time driving data and the historical driving data, calculating a driver error probability; performing risk grade division on the driver according to the driver error probability; executing a driving risk early warning strategy, judging whether the error probability of the driver exceeds an error probability threshold, and sending out an early warning signal; executing a signal lamp buffer adjustment strategy, and carrying out buffer adjustment on the yellow signal lamp period; the traffic jam is effectively relieved, and accidents are reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent transportation monitoring and early warning, and specifically provides an artificial intelligence-based urban intelligent transportation monitoring and early warning method, device, and medium. Background Art

[0002] With the increasing complexity of urban traffic, problems such as traffic congestion, frequent accidents, and unreasonable signal light cycles have gradually emerged. Traditional traffic management systems mainly rely on timed control and manual intervention. Although they can achieve traffic guidance to a certain extent, they cannot respond in real time to changes in traffic flow and dynamic changes in driver behavior, resulting in low traffic efficiency, high accident rates, especially during peak hours or under adverse weather conditions, where traffic flow prediction is inaccurate. With the development of artificial intelligence technology, especially the progress in fields such as image processing, deep learning, and traffic flow analysis, an artificial intelligence-based intelligent transportation monitoring and early warning method has gradually become an effective means to improve urban traffic management.

[0003] Existing intelligent traffic management systems still have limitations in many aspects. Many systems only rely on fixed traffic data inputs, such as vehicle speed and traffic flow, and fail to effectively utilize the combination of real-time monitoring images and weather data, ignoring the immediate impact of the changing environment on traffic flow. Current traffic signal cycle adjustments are mostly in a preset mode or based on single traffic flow data, and cannot dynamically optimize the yellow light cycle according to actual situations, resulting in increased traffic congestion or an elevated accident risk. Additionally, existing driver risk assessment methods are relatively simple and do not fully combine driver personal information and real-time behavior data, leading to inaccuracies in driving risk prediction, which in turn affects the effectiveness of traffic safety early warning. Existing strategies do not fully consider the reaction capabilities of different drivers. Especially for high-risk drivers, their reaction times may be longer and the likelihood of accidents is higher.

[0004] This solution integrates multiple advanced technologies, including real-time image processing, deep learning models, dynamic signal light cycle adjustment, and driver risk prediction. It can not only analyze traffic conditions in real time but also dynamically adjust traffic signal cycles according to road conditions, weather, and driver behavior, effectively alleviating traffic congestion and reducing accidents. Summary of the Invention

[0005] The present invention provides an artificial intelligence-based urban intelligent transportation monitoring and early warning method to help solve the problems mentioned in the above background art.

[0006] In a first aspect, the present application provides an artificial intelligence-based urban intelligent transportation monitoring and early warning method, adopting the following technical solution: An artificial intelligence-based urban intelligent transportation monitoring and early warning method includes:

[0007] S1. Obtain the traffic monitoring images collected by the monitoring at any signal intersection in the city in real time;

[0008] Screen and optimize the collected traffic monitoring images according to the collected traffic monitoring images;

[0009] S2. Generate a traffic condition prediction model and evaluate the impact of the current signal intersection information and weather information on traffic;

[0010] According to the historical traffic monitoring images, use image processing and deep learning models to identify abnormal behaviors in traffic flow and give early warnings;

[0011] S3. According to the real-time changes of traffic monitoring images and weather information, execute a dynamic adjustment strategy for the traffic signal cycle and adjust the cycle of traffic signals in real time;

[0012] S4. Obtain the driving information, historical driving data and real-time driving data of the driver;

[0013] Obtain all vehicle basic information and divide all vehicles into category one vehicles and category two vehicles;

[0014] Calculate the driver error probability according to the real-time driving data and historical driving data, combined with the driver's driving information;

[0015] S5. Set an error probability threshold to divide the risk levels of drivers;

[0016] According to the driver error probability, execute a driving risk warning strategy, judge whether the driver error probability exceeds the error probability threshold, and send a warning signal when the driver error probability exceeds the preset threshold;

[0017] S6. According to the risk levels of all drivers driving at the signal intersection, execute a signal light buffer adjustment strategy to buffer and adjust the yellow light signal cycle.

[0018] By obtaining and analyzing traffic monitoring images in real time, combining weather information and driver data, the intelligent level of urban traffic management can be effectively improved. The traffic condition prediction model can identify abnormal behaviors in time and give early warnings, so as to optimize the signal light cycle, reduce congestion and accident risks. At the same time, the driver risk assessment and warning mechanism can enhance driving safety, reduce the incidence of traffic accidents, and ultimately achieve a more efficient and safer urban traffic system.

[0019] Preferably, the screening and optimization of the collected traffic monitoring images according to the collected traffic monitoring images includes:

[0020] Collect traffic monitoring images in real time through road monitoring devices deployed at urban road nodes;

[0021] Extract the image clarity based on the acquired traffic monitoring images;

[0022] Set the image clarity standard, perform image clarity detection on the collected traffic monitoring images, and screen out the blurred images with image clarity less than the image clarity standard;

[0023] Adopt sharpening processing to repair the blurred images.

[0024] Through the comprehensive monitoring and image optimization of traffic intersections, improve the accuracy and reliability of traffic image data, provide a high-quality data basis for subsequent analysis, enhance the intelligence and real-time response ability of overall traffic management, help quickly detect traffic anomalies, and improve road traffic efficiency and the safety of urban traffic operation.

[0025] Preferably, based on historical traffic monitoring images, using image processing and deep learning models to identify abnormal behaviors in traffic flow and give early warnings, including:

[0026] Obtain historical traffic monitoring images and corresponding signalized intersection information, where the signalized intersection information includes traffic flow F, vehicle speed change v, congestion status C, accident occurrence records, and current weather information W. Traffic flow includes the number of vehicles n, vehicle speed, and traffic density on the road,

[0027] Based on historical traffic monitoring images, perform deep learning modeling to construct a traffic condition prediction model;

[0028] Set the training process of the traffic condition prediction model and introduce the weather factor at the corresponding moment as an auxiliary input variable;

[0029] Based on historical traffic monitoring images, use image processing algorithms and deep learning detection networks to real-time identify abnormal behaviors in traffic flow. The abnormal behaviors include vehicle reverse, frequent lane change, illegal occupation of road space, and running a red light;

[0030] When an abnormal behavior is detected, generate a warning message in time and record it. The warning message includes the type, location, and time of the abnormal behavior.

[0031] By using historical traffic monitoring images combined with image processing and deep learning models to identify traffic abnormal behaviors, it is possible to detect dangerous behaviors such as reverse, frequent lane change, illegal occupation of road space, and running a red light in advance, generate warning messages in time, which helps the traffic management department intervene and handle quickly, reduce the probability of traffic accidents, improve the road safety guarantee ability, and optimize the traffic circulation environment.

[0032] Preferably, based on the real-time changes of traffic monitoring images and weather information, execute a dynamic adjustment strategy for traffic signal cycle, and adjust the cycle of traffic signals in real time, including:

[0033] Obtain traffic monitoring images and weather information in real time to identify the current road environmental conditions;

[0034] Set the dynamic adjustment rules for traffic signal cycles, and establish a signal cycle adjustment model based on the current road environmental conditions;

[0035] Combined with historical traffic monitoring images, dynamically evaluate the current traffic operation situation S(t), S(t) = ω1·F + ω2·v + ω3·C + ω4·W, where ω1, ω2, ω3, ω4 represent weight coefficients;

[0036] Traffic flow F: Convert the number of vehicles n, vehicle speed, and traffic density into dimensionless values (range 0-1) through a normalization formula. (where n min / n max is the historical minimum / maximum traffic flow, v min / v max is the speed limit range, ρ min / ρ max is the historical minimum / maximum traffic density).

[0037] Vehicle speed change v: Take the percentage difference between the real-time vehicle speed and the section speed limit, and normalize it to the 0-1 interval.

[0038] Congestion condition C: Use the fuzzy logic algorithm to map it to 0 (unobstructed) - 1 (severely congested) according to parameters such as queue length and delay time.

[0039] Weather information W: Preset a weather level quantization table:

[0040]

[0041]

[0042] Obtain the initial duration of the signal lights, where the initial green light duration is T green and the initial yellow light duration is T yellow and the initial red light duration is T red ;

[0043] Adjust according to the initial green light duration, initial yellow light duration, initial red light duration and phase sequence switching logic:

[0044] T green0 = T green + α1·S(t) + β1·f(W)

[0045] T yellow0 = T yellow + α2·S(t)

[0046] T red0 = T red + α3·S(t) - β2·f(W)

[0047] Wherein, T green0 , T yellow0 , T red0 , are respectively the initial green light duration after adjustment, the initial yellow light duration after adjustment, and the initial red light duration after adjustment. α1, α2, α3 are road condition adjustment coefficients, β1, β2 are weather adjustment coefficients, and f(W) represents the influence function of weather factors on the signal light duration.

[0048] Correction coefficient for adjusting signal light duration according to weather grade: f(W) = 1 - 0.5×(1 - W), for green light / red light adjustment, f(W) <= 0.2×(1 < - < W), for yellow light adjustment.

[0049] By dynamically adjusting the traffic signal light cycle according to traffic monitoring images and weather changes, the traffic light timing can be flexibly optimized according to real-time road conditions, congestion can be alleviated, and traffic efficiency can be improved. In case of bad weather or emergencies, it can quickly adapt to changes, effectively avoid abnormal accumulation of traffic flow, and improve the self-adaptability and intelligent response level of the urban traffic system.

[0050] Preferably, calculating the driver error probability according to real-time driving data and historical driving data, combined with the driver's driving information, includes:

[0051] Obtaining the driver's driving information and historical driving data through traffic monitoring images. The driving information includes age A, gender, driving experience D, and driver's license category. The historical driving records include violation records, accident records, and driving habits H;

[0052] Obtaining the basic information of all vehicles and classifying all vehicles into class one vehicles V1 and class two vehicles V2. Among them, class one vehicles are vehicles with a practice sign, and class two vehicles are other vehicles without a practice sign;

[0053] · According to real-time driving data R and historical driving data, focusing on the driving conditions of class one vehicles, combined with the driver's driving information, analyzing and calculating the error probability P of any driver, Wherein, γ1, γ2, γ3, γ4, γ5 are the weight coefficients of the error probability, A max and D max represent the maximum limits of age and driving experience.

[0054] Historical driving habit score H, establishing a driving habit scoring system, calculated based on data such as historical violation records, accident frequencies, and the number of hard brakes / hard turns: (The result is normalized to 0 - 1, and the higher the value, the higher the habit risk).

[0055] Conversion of a type of vehicle V1

[0056] Set a fixed risk coefficient for a type of vehicle (learner driver's vehicle) V1 <= < 1, and for a type of vehicle V2 <= < 0, and directly substitute them into the formula as Boolean variables.

[0057] Real-time driving data risk score R

[0058] Collect behavior data such as hard acceleration, hard braking, and lane departure in real time through in-vehicle sensors, and use a support vector machine (SVM) model to output a risk score R (range < 0 - 1).

[0059] Model training data: Samples in historical driving data marked with high-risk behaviors (such as data in the 10 minutes before an accident).

[0060] By combining real-time driving data, historical driving data, and driver's basic information, calculate the driver's error probability, which helps to accurately identify potential high-risk driver groups. Pay special attention to special vehicle categories such as learner drivers, improve the pertinence and preventive nature of traffic management, master the risk situation in advance, and reduce potential traffic accident hazards caused by human errors.

[0061] Preferably, when implementing the driving risk warning strategy, set an error probability threshold, and judge whether the driver's error probability exceeds the error probability threshold, including:

[0062] Set an error probability threshold range ΔP1, ΔP2;

[0063] According to the driver's error probability P, divide the risk level of the driver's driving;

[0064] When P > ΔP1, the driver is classified as a high-risk level N1;

[0065] When ΔP2 < P ≤ ΔP1, the driver is classified as a medium-risk level N2;

[0066] When P ≤ ΔP2, the driver is classified as a low-risk level N3, where N1 > N2 > N3;

[0067] When the driving risk reaches the medium-risk level N2, an early warning signal is immediately issued;

[0068] If a driver at the high-risk level N1 is monitored, then monitor the driver's real-time driving data in real time and continuously update the risk level.

[0069] By setting the driver error probability threshold and dividing the risk levels, the quantitative management of driving behavior risks can be achieved, and early warning and intervention for medium- and high-risk drivers can be carried out in a timely manner to prevent the further deterioration of dangerous behaviors. Dynamically updating the driving risk levels is conducive to building a more intelligent and dynamic traffic safety supervision system and improving the overall road traffic safety level.

[0070] Preferably, according to the risk levels of all drivers driving at the signalized intersection, the signal buffer adjustment strategy is executed to buffer and adjust the yellow light signal cycle, including:

[0071] Set the vehicle number threshold Δn at the signalized intersection;

[0072] According to the risk levels of all drivers driving at any signalized intersection, when the vehicle number n at this signalized intersection exceeds Δn, execute the signal buffer adjustment strategy;

[0073] Calculate the overall risk level N of this intersection z , where N i represents the risk level of the driver of the i-th vehicle at this signalized intersection, represents the weight coefficient of the i-th vehicle, and U represents the total weight;

[0074] Set the maximum duration of the yellow light signal cycle as T max ;

[0075] When N3 ≤ N z < N2, adjust the cycle of the yellow light signal to T′ yellow3 = min(T yellow ×(1 + k3), T max ), where k3 is the low-risk gain coefficient;

[0076] When N2 ≤ N z < N1, adjust the cycle of the yellow light signal to T′ yellow2 = min(T yellow ×(1 + k2), T max ), where k2 is the medium-risk gain coefficient;

[0077] When N z ≥ N1, adjust the cycle of the yellow light signal to T′ yellow1 = min(T yellow ×(1 + k1), T max ), where k1 is the high-risk gain coefficient.

[0078] By implementing a yellow light buffer adjustment strategy based on the overall risk level of drivers at signalized intersections, it is possible to dynamically optimize the yellow light duration according to the potential risks in the traffic flow, give drivers more reaction time, reduce the behavior of running through red lights or sudden braking caused by insufficient yellow light duration, improve the traffic safety at intersections, and further reduce the incidence of traffic accidents.

[0079] In a second aspect, the present application provides an artificial intelligence-based urban intelligent traffic monitoring and warning device, adopting the following technical solution: An artificial intelligence-based urban intelligent traffic monitoring and warning device includes:

[0080] A collection module: used to deploy monitoring devices at the main road nodes in the city, collect traffic monitoring image data in real time, collect the basic information of drivers, historical driving data and real-time driving behaviors, and collect the basic information of all vehicles, providing basic data for driving risk assessment;

[0081] An abnormal behavior detection module: used to use image processing algorithms and deep learning detection networks to identify abnormal behaviors in the traffic flow in real time;

[0082] A signal light cycle adjustment module: used to dynamically adjust the cycle and phase sequence switching logic of the green, yellow, and red lights of the traffic signal according to the traffic monitoring images and weather changes;

[0083] An error probability calculation module: used to calculate the error probability of the driver according to the driver's age, gender, driving experience, driver's license category, driving habits and driving records, combined with real-time driving data;

[0084] A driving risk warning module: used to set a risk level threshold according to the error probability, classify the risk level of the driver, and send a warning signal in time when the medium to high risk level is reached;

[0085] A yellow light signal buffer module: used to intelligently and dynamically adjust the cycle of the yellow light signal according to the calculation result of the comprehensive risk level of the intersection, so as to relieve the traffic pressure in high-risk situations and improve traffic safety and fluency.

[0086] The present invention has the following beneficial effects:

[0087] 1. This is an artificial intelligence urban intelligent traffic monitoring and early warning method. By introducing an artificial intelligence urban intelligent traffic monitoring and early warning method, urban traffic management will achieve all-round intelligence and automation. This technology can acquire and analyze traffic monitoring images of any signal intersection in the city in real time to ensure timely understanding and response to traffic conditions. By optimizing the monitoring images and improving the image quality, it can more accurately capture traffic flow and abnormal behaviors. Once potential problems are identified, such as driving in the wrong direction, running red lights, etc., the system will quickly issue early warning information to ensure that relevant departments can take timely measures, thereby effectively reducing the accident rate and traffic congestion, and improving the traffic efficiency of the entire city.

[0088] 2. This is an artificial intelligence-based urban intelligent traffic monitoring and early warning method. By establishing a traffic condition prediction model and combining real-time weather and intersection information, the dynamic adjustment mechanism of traffic lights can respond flexibly according to actual conditions. This dynamic adjustment is not only based on historical data analysis, but also takes into account real-time traffic flow and environmental changes, so as to reasonably set the duration of the green, yellow, and red lights of the traffic lights. This intelligent adjustment of the traffic light cycle can effectively reduce traffic bottlenecks, improve traffic efficiency, save road traffic time, and thus improve the travel experience of citizens. At the same time, real-time monitoring of traffic flow also provides more accurate data support for subsequent traffic planning and management, making traffic management more scientific and efficient.

[0089] 3. This artificial intelligence urban intelligent traffic monitoring and early warning method, through real-time monitoring and evaluation of the driver's driving behavior, the system can not only timely identify high-risk driving behavior, but also accurately divide the risk level according to the driver's historical data and driving habits. This risk-based management method enables traffic management agencies to focus on and monitor high-risk drivers in real time, effectively preventing the occurrence of traffic accidents. At the same time, by setting a threshold for the probability of error, the system can actively issue an early warning when the driver's risk level reaches medium or high risk, reminding the driver to pay attention to safety. By increasing the length of the yellow light to improve the reaction time of high-risk drivers, the defects of existing technologies can be effectively compensated. This strategy not only helps to reduce the possibility of accidents, but also improves overall traffic safety and efficiency. Such an early warning mechanism not only improves driving safety, but also promotes the driver's safety awareness to a certain extent, forms a good traffic safety culture, and ultimately achieves a safe and efficient urban traffic environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0090] Figure 1 It is a schematic diagram of the process of the present invention.

[0091] Figure 2 It is a schematic diagram of the structure of the device of the present invention. DETAILED DESCRIPTION

[0092] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0093] Example 1, referring to Figure 1 , an urban intelligent transportation monitoring and early warning method for artificial intelligence, including:

[0094] S1. Real-time obtain traffic monitoring images collected by monitoring at any signal intersection in the city, and screen and optimize the collected traffic monitoring images according to the collected traffic monitoring images, including:

[0095] High-definition monitoring camera devices are deployed at the intersections of the main roads in the city to collect traffic monitoring images in real time, and the image resolution is 1920×1080;

[0096] Set the image clarity standard value to 80 points. If the clarity of the collected image is lower than 80 points, it is determined as a blurred image, and the blurred image is repaired by sharpening filtering to make the repaired image reach 90 points.

[0097] By setting the image clarity standard and performing sharpening repair on the blurred image, the accuracy of subsequent image processing and analysis can be significantly improved, and the recognition misjudgment caused by image quality problems can be reduced, providing a solid image foundation for subsequent intelligent traffic management.

[0098] S2. According to the historical traffic monitoring images, use image processing and deep learning models to identify abnormal behaviors in traffic flow and give early warnings, including:

[0099] Collect historical traffic monitoring images of 500 main signal intersections in a certain city throughout 2024, with a total of about 3 million images, and extract the following data:

[0100] Traffic flow: The number of vehicles per minute is between 20 and 200;

[0101] Vehicle speed change: The normal range is 20 km / h to 60 km / h;

[0102] Congestion status C: Quantified by a congestion coefficient of 0 to 1 (0 means unobstructed, 1 means severely congested);

[0103] Weather information W: Divided into sunny day (1), rainy day (0.7), and snowy day (0.5).

[0104] A traffic prediction model is established through a deep convolutional neural network to identify abnormal behaviors, including: driving in the wrong direction (with an accuracy of 98%); frequent lane changes (with an accuracy of 95%); illegal road occupation (with an accuracy of 93%); and running red lights (with an accuracy of 96%).

[0105] When an abnormality is detected, the system generates a warning message within 0.5 seconds and sends it to the traffic management platform.

[0106] By introducing deep convolutional neural networks to identify abnormal traffic behaviors, it can not only achieve high-precision identification, but also greatly improve the early warning response speed to potential risks, reduce human intervention, and improve the intelligent level of urban traffic safety management. At the same time, the model is trained with large-scale historical image data and has good generalization ability, which is suitable for different cities and different road conditions.

[0107] S3. According to the real-time changes of traffic monitoring images and weather information, a traffic light cycle strategy is dynamically adjusted to adjust the traffic light cycle in real time, including:

[0108] Obtain traffic monitoring images and weather information in real time to identify current road environment conditions;

[0109] Get the initial green light duration T green = 60 seconds; initial yellow light duration T yellow = 3 seconds; initial red light duration T red = 60 seconds;

[0110] The current road condition adjustment coefficient is S(t) = 0.7, where 1 means smooth traffic and 0 means congestion. At this time, the weather is light rain, and f(W) = 0.85;

[0111] Set the road condition adjustment coefficients to α1=-10, α2=1, α3=10; the weather adjustment coefficients to β1=-5, β2=5;

[0112] The adjusted initial duration of the green light is T green0 =T green +α1·S(t)+β1·f(W)=60+(-10)×0.7+(-5)×0.85=48.75 seconds;

[0113] The initial duration of the yellow light after adjustment is T yellow0 =T yellow +α2·S(t)=3+1×0.7=3.7 seconds;

[0114] The initial duration of the red light after adjustment is T red0 =T red +α3·S(t)-β2·f(W)=60+10×0.7-5×0.85=62.75 seconds.

[0115] By combining real-time traffic images with weather information and dynamically adjusting the signal light cycle, it is possible to more precisely adapt to current road conditions, reduce unnecessary waiting time, improve vehicle passing efficiency, and alleviate traffic congestion. It also has the ability to quickly respond to emergencies, which helps to enhance the flexibility and timeliness of traffic management.

[0116] S4. Calculate the driver error probability based on real-time driving data and historical driving data, combined with the driver's driving information, including:

[0117] Obtain the information of a driver as follows: age A = 22 years old, driving experience D = 1 year, driver's license category is C1, historical driving habit score H = 0.6 points (ranging from 0 to 1, the closer to 1, the more dangerous the habit), the vehicle is a type of vehicle V1, and the risk score of the actual driving data R = 0.7 (ranging from 0 to 1, the closer to 1, the higher the risk);

[0118] Set the maximum age A max = 70 years old, the maximum driving experience D max = 50 years, and the weight coefficients of the error probability are γ1 = 0.2, γ2 = 0.15, γ3 = 0.25, γ4 = 0.2, γ5 = 0.2;

[0119] Calculate the error probability

[0120] By integrating multi-dimensional information such as personal driving behavior data, age, driving experience, and vehicle type, an error probability model is constructed, which can more scientifically and personalizedly evaluate driving risks, provide data support for refined traffic safety management, and also provide a quantifiable reference for the subsequent early warning mechanism.

[0121] S5. Execute the driving risk early warning strategy according to the driver error probability, and determine whether the driver error probability exceeds the error probability threshold. When the driver error probability exceeds the preset threshold, an early warning signal is issued, including:

[0122] Set the error probability division thresholds as ΔP1 = 0.6 and ΔP2 = 0.3;

[0123] When P > ΔP1, the driver is classified as a high-risk level N1;

[0124] When ΔP2 < P ≤ ΔP1, the driver is classified as a medium-risk level N2;

[0125] When P ≤ ΔP2, the driver is classified as a low-risk level N3, where N1 > N2 > N3;

[0126] According to the driver's error probability ΔP2 = 0.3 < P = 0.55586 ≤ ΔP1 = 0.6, the risk level of this driver is classified as the medium risk level N2.

[0127] By establishing a hierarchical early warning mechanism based on the error probability, targeted management and intervention can be carried out for drivers with different risk levels, and the occurrence of high-risk behaviors can be avoided in advance. Especially the accurate identification of high-risk drivers can effectively reduce the traffic accident rate and improve the safety of the overall traffic system.

[0128] S6. According to the risk levels of all drivers driving at the signalized intersection, implement the signal buffer adjustment strategy to buffer and adjust the yellow light signal cycle, including:

[0129] Since no matter how many signal lights there are at the signalized intersection, the duration of the yellow light is not the main factor for adjusting the traffic conditions at the intersection, that is, by adjusting the duration of the yellow light, no traffic conflicts will be caused. When there are multiple drivers with high risk levels driving vehicles at this intersection, appropriately increasing the duration of the yellow light can increase the reaction time of the drivers, improve the reaction accuracy, and reduce the possibility of accidents;

[0130] Set the vehicle quantity threshold Δn = 10 vehicles at the signalized intersection. At a certain moment, the total weight is considered, and the vehicle quantity n = 15 vehicles is monitored at this intersection. Set the error probabilities of vehicle 1 - vehicle 15 as follows:

[0131] N′1 = 0.2, N′2 = 0.5, N′3 = 0.4, N′4 = 0.7, N′5 = 0.3, N′6 = 0.1,, N′7 = 0.6, N′8 = 0.8, N′9 = 0.2, N′ 10 = 0.4, N′ 11 = 0.5, N′ 12 = 0.3, N′ 13 = 0.1, N′ 14 = 0.9, N′ 15 = 0.4;

[0132] Calculate the overall error probability N of this intersection z = 0.47,

[0133] Set the maximum duration of the yellow light signal cycle as T max = 5, and the medium risk gain coefficient k2 = 0.5;

[0134] N z = <0.47 belongs to the medium risk level, and adjust the cycle of the yellow light signal to T′ yellow2 = min(T yellow ×(1 + k2), T max ) = min(3×(1 + 0.5), 5) = 4.5 seconds.

[0135] By adjusting the yellow light cycle based on the real-time driver risk level at intersections, the reaction time and operation margin of drivers can be effectively improved without affecting the overall traffic efficiency, and the adaptability of the traffic signal system to risk situations can be enhanced. This buffering mechanism provides a safety redundancy design, which is beneficial to reducing the collision risk caused by insufficient emergency response.

[0136] Example 2, referring to Figure 2 , an urban intelligent transportation monitoring and warning method of artificial intelligence, including:

[0137] Collection module: used to deploy monitoring devices at major road nodes in the city, collect traffic monitoring image data in real time, collect basic information of drivers, historical driving data and real-time driving behaviors, and collect basic information of all vehicles to provide basic data for driving risk assessment;

[0138] Abnormal behavior detection module: used to use image processing algorithms and deep learning detection networks to identify abnormal behaviors in traffic flow in real time;

[0139] Signal light cycle adjustment module: used to dynamically adjust the cycle and phase sequence switching logic of green lights, yellow lights, and red lights of traffic signals according to traffic monitoring images and weather changes;

[0140] Error probability calculation module: used to calculate the error probability of drivers according to the age, gender, driving age, driver's license category, driving habits and driving records of drivers, combined with real-time driving data;

[0141] Driving risk warning module: used to set risk level thresholds according to error probabilities, classify the risk levels of drivers, and issue warning signals in a timely manner when medium to high risks are reached;

[0142] Yellow light signal buffer module: used to intelligently and dynamically adjust the cycle of yellow light signals according to the calculation results of the comprehensive risk level at intersections to relieve traffic pressure in high-risk situations and improve traffic safety and smoothness.

[0143] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device.

[0144] The above are only the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the technical principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. An urban intelligent transportation monitoring and early warning method based on artificial intelligence, characterized in that, Including: S1. Obtain traffic monitoring images collected by monitoring at any signal intersection in the city in real time; Screen and optimize the collected traffic monitoring images according to the collected traffic monitoring images; S2. Generate a traffic condition prediction model and evaluate the impact of the current signal intersection information and weather information on traffic; According to historical traffic monitoring images, use image processing and deep learning models to identify abnormal behaviors in traffic flow and give early warnings; S3. According to the real-time changes of traffic monitoring images and weather information, execute a dynamic traffic signal cycle adjustment strategy to adjust the cycle of traffic signals in real time; S4. Obtain the driving information, historical driving data and real-time driving data of the driver; Obtain all vehicle basic information and divide all vehicles into Class I vehicles and Class II vehicles; Calculate the driver error probability according to the real-time driving data and historical driving data, combined with the driver's driving information; S5. Set an error probability threshold to divide the driver's risk level; According to the driver error probability, execute a driving risk warning strategy to judge whether the driver error probability exceeds the error probability threshold, and send a warning signal when the driver error probability exceeds the preset threshold; S6. According to the risk levels of all drivers at the signal intersection, execute a signal light buffer adjustment strategy to buffer and adjust the yellow light signal cycle.

2. The urban intelligent transportation monitoring and early warning method of artificial intelligence according to claim 1, characterized in that, The screening and optimization of the collected traffic monitoring images according to the collected traffic monitoring images includes: Collect traffic monitoring images in real time through road monitoring devices deployed at urban road nodes; Extract the image clarity according to the obtained traffic monitoring images; Set an image clarity standard, detect the image clarity of the collected traffic monitoring images, and screen out blurred images with image clarity less than the image clarity standard; Use sharpening processing to repair the blurred images.

3. An urban intelligent transportation monitoring and early warning method for artificial intelligence according to claim 1, characterized in that, The identification of abnormal behaviors in traffic flow and giving early warnings according to historical traffic monitoring images, using image processing and deep learning models includes: Obtain historical traffic monitoring images and corresponding signal intersection information, where the signal intersection information includes traffic flow F, vehicle speed change v, congestion condition C, accident occurrence records and the weather information W at that time, and the traffic flow includes the number of vehicles n, vehicle speed and traffic density on the road; Execute deep learning modeling according to historical traffic monitoring images to build a traffic condition prediction model; Set the training process of the traffic condition prediction model and introduce the weather factors at the corresponding moment as auxiliary input variables; According to historical traffic monitoring images, use image processing algorithms and deep learning detection networks to identify abnormal behaviors in traffic flow in real time. The abnormal behaviors include vehicle reverse, frequent lane change, illegal occupation of road space, and running a red light; When abnormal behaviors are detected, generate and record warning information, where the warning information includes the type, location and time of the abnormal behavior.

4. An urban intelligent transportation monitoring and early warning method for artificial intelligence according to claim 3, characterized in that, The execution of the dynamic traffic signal cycle adjustment strategy to adjust the cycle of traffic signals in real time according to the real-time changes of traffic monitoring images and weather information includes: Obtain traffic monitoring images and weather information in real time to identify the current road environment conditions; Set the dynamic adjustment rules for the traffic signal cycle, and establish a signal cycle adjustment model according to the current road environment conditions; Combined with historical traffic monitoring images, dynamically evaluate the current traffic operation situation S(t), S(t) = ω1·F + ω2·v + ω3·C + ω4·W, where ω1, ω2, ω3, ω4 represent weight coefficients; Obtain the initial duration of the signal light, where the initial duration of the green light is T green , the initial duration of the yellow light is T yellow , the initial duration of the red light is T red ; Adjust according to the initial green light duration, initial yellow light duration, initial red light duration and phase sequence switching logic: T green0 = T green + α1·S(t) + β1·f(W) T yellow0 = T yellow + α2·S(t) T red0 = T red + α3·S(t) - β2·f(W) Among them, T green0 , T yellow0 , T red0 are respectively the initial green light duration after adjustment, the initial yellow light duration after adjustment, and the initial red light duration after adjustment. α1, α2, α3 are road condition adjustment coefficients, β1, β2 are weather adjustment coefficients, and f(W) represents the influence function of weather factors on the signal light duration.

5. An urban intelligent transportation monitoring and early warning method for artificial intelligence according to claim 1, characterized in that, The calculation of the driver's error probability according to real-time driving data and historical driving data, combined with the driver's driving information, includes: Obtain the driver's driving information and historical driving data through traffic monitoring images. The driving information includes age A, gender, driving experience D and driver's license category. The historical driving records include violation records, accident records and driving habits H; Obtain the basic information of all vehicles, and divide all vehicles into category 1 vehicles V1 and category 2 vehicles V2. Among them, category 1 vehicles are vehicles with a novice sign, and category 2 vehicles are other vehicles without a novice sign; Based on real-time driving data R and historical driving data, focus on the driving conditions of a certain type of vehicle, and combine the driving information of the driver to analyze and calculate the error probability P of any driver. Among them, Y1, γ2, Y3, γ4, γ5 are the weight coefficients of the error probability, A max and D max represent the maximum limits of age and driving experience.

6. The urban intelligent transportation monitoring and early warning method of artificial intelligence according to claim 5, characterized in that, Execute the driving risk warning strategy, set the error probability threshold, and judge whether the driver's error probability exceeds the error probability threshold, including: Set the error probability threshold range ΔP1, ΔP2; Divide the risk level of the driver's driving according to the driver's error probability P; When P > ΔP1, the driver is classified as a high risk level N1; When ΔP2 < P ≤ ΔP1, the driver is classified as a medium risk level N2; When P ≤ ΔP2, the driver is classified as a low risk level N3, where N1 > N2 > N3; When the driving risk reaches the medium risk level N2, an alarm signal is immediately issued; If a driver with a high risk level N1 is detected, the real-time driving data of the driver is monitored in real time, and the risk level is continuously updated.

7. An urban intelligent transportation monitoring and warning method for artificial intelligence according to claim 6, characterized in that, Execute the signal lamp buffer adjustment strategy according to the risk levels of all drivers at the signal intersection, and buffer-adjust the yellow light signal cycle, including: Set the vehicle number threshold Δn at the signal intersection; According to the risk levels of all drivers at any signal intersection, when the vehicle number n at the signal intersection exceeds Δn, execute the signal lamp buffer adjustment strategy; Calculate the overall risk level N of this intersection z , where N i represents the risk level of the driver of the i-th vehicle at this signalized intersection, represents the weight coefficient of the i-th vehicle, and U represents the total weight; Set the maximum duration of the yellow light signal cycle to T max ; When N3 ≤ N z <When N2, adjust the cycle of the yellow - light signal to T′ yellow3 = min(T yellow × (1 + k3), T max ), where k3 is the low - risk gain coefficient; When N2 ≤ N z <When N1, adjust the cycle of the yellow traffic light to T' yellow2 =min(T yellow ×(1 + k2), T max ), where k2 is the medium-risk gain coefficient; When N z ≥ N1, adjust the cycle of the yellow - light signal to T′ yellow1 = min(T yellow × (1 + k1), T max ), where k1 is the high - risk gain coefficient.

8. An urban intelligent transportation monitoring and early warning device for artificial intelligence, which is applied to the urban intelligent transportation monitoring and early warning method for artificial intelligence described in any one of claims 1-7, and is characterized in that, Including: Collection module: Used to deploy monitoring devices at the main road nodes in the city, collect traffic monitoring image data in real time, collect the basic information of drivers, historical driving data and real-time driving behaviors, collect the basic information of all vehicles, and provide basic data for driving risk assessment; Abnormal behavior detection module: Used to use image processing algorithms and deep learning detection networks to real-time identify abnormal behaviors in traffic flow; Signal lamp cycle adjustment module: Used to dynamically adjust the cycle and phase sequence switching logic of green lights, yellow lights and red lights of traffic signals according to traffic monitoring images and weather changes; Error probability calculation module: Used to calculate the driver's error probability according to the driver's age, gender, driving experience, driver's license category, driving habits and driving records, combined with real-time driving data; Driving risk warning module: used to set the risk level threshold according to the error probability, classify the driver's risk level, and issue a warning signal in a timely manner when the medium to high risk level is reached; Yellow light signal buffer module: used to intelligently and dynamically adjust the cycle of the yellow light signal according to the calculation result of the comprehensive risk level of the intersection, so as to relieve the traffic pressure in high-risk situations and improve traffic safety and fluency.

9. A computer-readable storage medium, characterized in that, A computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the steps of an artificial intelligence-based urban intelligent traffic monitoring and warning method according to any one of claims 1 to 8 are implemented.

Images