Urban road network and mobile communication network fused traffic risk early warning method and system

Abstract

The invention provides an urban road network and mobile communication network fused traffic risk early warning method and system, and belongs to the technical field of traffic risk early warning. Comprising the following steps: S1, constructing a handling risk data set; s2, constructing a spatial-temporal feature fusion model by using a convolutional neural network; s3, constructing a vehicle abnormal behavior coefficient, a crowd abnormal behavior coefficient and a personnel growth index of the ith street and the jth monitoring area based on the traffic and governance risk data set; and S4, constructing a human-vehicle abnormal ratio coefficient, presetting an abnormal threshold value, when the human-vehicle abnormal ratio coefficient is greater than 1, determining that the event is a public security type event, and when the human-vehicle abnormal ratio coefficient is less than or equal to 1, determining that the event is a traffic type event. According to the method, traffic flow dynamic characteristics and a crowd time sequence change rule are comprehensively considered, a risk change mode can be automatically learned in a multi-dimensional characteristic space, and compared with a traditional rule threshold value judgment mode, the method has higher risk prediction precision and response speed.

Description

Technical Field

[0001] This invention relates to the field of traffic risk early warning technology, and more specifically, to a traffic risk early warning method and system that integrates urban road networks and mobile communication networks. Background Technology

[0002] As urbanization continues, traffic flow and crowd activity on city streets are becoming increasingly dynamic and complex. In nighttime commercial areas, densely populated residential areas, and key public areas, there has been a significant increase in frequent vehicle entry and exit, illegal parking, and crowd gatherings, putting considerable pressure on traffic order and public security management.

[0003] In response to this situation, most existing street monitoring methods rely on manual patrols, which cannot fully utilize the spatiotemporal dynamic characteristics of vehicles and crowds. This leads to delayed event warnings and low efficiency in resource allocation. Furthermore, traffic management and public security management systems are often isolated and lack cross-system data fusion mechanisms, making it difficult to make comprehensive judgments on emergencies. For example, when a fight breaks out in a bar street at night, traditional systems may not be able to quickly determine the connection between crowd gathering and surrounding traffic congestion, nor can they promptly dispatch police officers and traffic police to coordinate the handling of the situation. Summary of the Invention

[0004] To overcome the above deficiencies, the present invention provides a traffic risk early warning method and system for the integration of urban road networks and mobile communication networks to overcome or at least partially solve the above technical problems.

[0005] This invention is implemented as follows: This invention provides a security risk early warning method integrating urban road networks and mobile communication networks, comprising: S1. Based on the traffic sensing terminal deployed in the j-th monitoring area of ​​the i-th street in the urban road network, and the mobile communication base station covering the j-th monitoring area of ​​the i-th street, collect the frequency of repeated vehicle entry and exit, the number of vehicles staying in prohibited parking locations, the number of vehicles, the speed of pedestrian flow gathering, and the number of people in the j-th monitoring area of ​​the i-th street, and construct a traffic management risk dataset. S2. Using a convolutional neural network, construct a spatiotemporal feature fusion model, input the traffic risk dataset into the spatiotemporal feature fusion model, and output the risk prediction results; S3. Based on the traffic risk dataset, construct the abnormal vehicle behavior coefficients for the j-th monitoring area on the i-th street. Population Abnormal Behavior Coefficient and personnel growth index ; S4. Vehicle abnormal behavior coefficients based on the j-th monitoring area of ​​the i-th street. Population Abnormal Behavior Coefficient and personnel growth index Combined, construct the ratio coefficient of human and vehicle anomalies. And preset anomaly thresholds, when the ratio of people and vehicles to anomalies is... When the ratio is greater than 1, it is determined to be a public security incident; when the ratio of the abnormal ratio of people and vehicles is less than or equal to 1, it is determined to be a traffic incident.

[0006] In a preferred embodiment, S1 includes; S11. By installing a multi-view high-definition camera in the j-th monitoring area on the i-th street, vehicle images in the j-th monitoring area are acquired using the high-definition camera and transmitted to an image analysis terminal via mobile communication. After noise reduction processing, vehicle textures in the images are identified to obtain the number of vehicles in the j-th monitoring area. An image recognition algorithm is then used to identify the license plate of each vehicle in the j-th monitoring area, and the frequency of repeated entry and exit of vehicles in the j-th monitoring area on the i-th street is monitored. ; S12. Take images of the no-parking locations in the j-th monitoring area using a high-definition camera, and transmit them to an image analysis terminal via mobile communication. Identify vehicle texture features in the images and obtain the number of vehicles staying at the no-parking locations in the j-th monitoring area on the i-th street. .

[0007] In a preferred embodiment, S1 further includes; S13. Based on the high-definition camera covering the j-th monitoring area of ​​the i-th street, capture images of the j-th monitoring area in real time, and transmit them to the image analysis terminal via mobile communication. Enhance the images, use a human target detection algorithm to identify the number of people in the images, and obtain the number of people in the j-th monitoring area of ​​the i-th street. And collect the number of people in the j-th monitoring area of ​​the i-th street per unit time. Calculate the pedestrian flow aggregation speed in the j-th monitoring area of ​​the i-th street. This is used to construct a traffic risk dataset.

[0008] In a preferred embodiment, S2 includes; S21. Construct a spatiotemporal feature fusion model using a convolutional neural network, train and test the spatiotemporal feature fusion model with a traffic risk dataset, and use the trained spatiotemporal feature fusion model as a traffic risk test and evaluation model. At the same time, use the intermediate layer output of the device running spatiotemporal feature fusion model as a feature vector to identify feature information, and use the trained spatiotemporal feature fusion model as a data running prediction.

[0009] In a preferred embodiment, S3 includes; S31. Frequency of repeated vehicle entry and exit in the j-th monitoring area of ​​the i-th street based on the traffic risk dataset. Number of vehicles staying in no-parking zones Number of vehicles The abnormal vehicle behavior coefficient of the j-th monitoring area on the i-th street is calculated using the following method. ; First, the frequency of vehicle entry and exit in the j-th monitoring area of ​​the i-th street is repeatedly counted. Number of vehicles staying in no-parking zones Number of vehicles After normalization, the frequency of repeated vehicle entry and exit in the j-th monitoring area of ​​the i-th street is obtained. Number of vehicles staying in no-parking zones Number of vehicles ; In the formula, This represents the maximum frequency of repeated vehicle entry and exit. In the formula, This represents the maximum number of vehicles that can remain in a no-parking zone. In the formula, This represents the maximum number of vehicles. Secondly, based on the normalized frequency of repeated vehicle entry and exit in the j-th monitoring area of ​​the i-th street. Number of vehicles staying in no-parking zones Number of vehicles The abnormal vehicle behavior coefficient of the j-th monitoring area on the i-th street is calculated using the following formula. ; ; In the formula , and This represents the weighting coefficient.

[0010] In a preferred embodiment, S3 further includes; S32. By setting a preset abnormal vehicle behavior threshold A, the abnormal vehicle behavior coefficient of the j-th monitoring area on the i-th street is... Compare with the vehicle abnormal behavior threshold A to generate a vehicle abnormal behavior assessment instruction, including: when When >A, it indicates that the vehicle behavior in the j-th monitoring area of ​​the i-th street is abnormal, and a vehicle abnormality alarm instruction is sent to the traffic management terminal to dispatch nearby traffic police to handle the situation and direct traffic. when When A ≤ A, it means that the vehicle behavior in the j-th monitoring area of ​​the i-th street is normal.

[0011] In a preferred embodiment, S3 further includes; S33. Population size in the j-th monitoring area of ​​the i-th street based on the traffic risk dataset. The abnormal behavior coefficients of the population in the j-th monitoring area of ​​the i-th street are obtained through the following method. ; In the formula Expressed as a unit of time, It represents the historical average number of people in the j-th monitoring area of ​​the i-th street within a unit of time. It is expressed as the historical standard deviation of the population size in the j-th monitoring area of ​​the i-th street within a unit of time. S34. By setting a threshold S for abnormal crowd behavior, the coefficient of abnormal crowd behavior in the j-th monitoring area of ​​the i-th street is... Compare with the abnormal behavior threshold S of the population to generate an abnormal behavior assessment instruction for the population, including: when When the value is greater than S, it indicates that the number of people gathering in the j-th monitoring area of ​​the i-th street is abnormal, and an abnormal alarm instruction is sent to the public security command terminal, requiring nearby police officers to be dispatched to evacuate the crowd. when When ≤S, it means that the number of people in the j-th monitoring area of ​​the i-th street is normal.

[0012] In a preferred embodiment, S3 further includes; S35. Population count in the j-th monitoring area of ​​the i-th street. The pedestrian flow aggregation speed in the j-th monitoring area of ​​the i-th street is calculated using the following formula. ; In the formula, The time interval is represented by the number of people collected. Subsequently, based on the pedestrian flow aggregation speed in the j-th monitoring area of ​​the i-th street... The population growth index of the j-th monitoring area on the i-th street is calculated using the following formula. ; In the formula, It is represented as a time span factor.

[0013] In a preferred embodiment, S4 includes; S41. Based on the abnormal vehicle behavior coefficient of the j-th monitoring area on the i-th street. Population Abnormal Behavior Coefficient and personnel growth index Combined, the ratio coefficient of pedestrian and vehicle anomalies in the j-th monitoring area of ​​the i-th street is calculated using the following formula. ; First, calculate the abnormal vehicle behavior coefficient for the j-th monitoring area on the i-th street. By performing quantile mapping according to the historical sample distribution, standardized parameters of vehicle anomalies are obtained. ; The abnormal behavior coefficient of the population in the j-th monitoring area of ​​the i-th street. Mapped to standardized parameters ; The population growth index of the j-th monitoring area on the i-th street The normalized growth index of personnel was obtained by comparing with the historical 95th percentile. ; Then, the pedestrian-vehicle anomaly ratio coefficient for the j-th monitoring area on the i-th street is calculated using the following formula. ; In the formula Represented as a natural number divisible by zero, with a value of 0.001; S42. Calculate the ratio of pedestrian and vehicle anomalies in the j-th monitoring area of ​​the i-th street. Compare with 1; when When the value is greater than 1, it indicates that a public security incident has occurred in the j-th monitoring area of ​​the i-th street, requiring the dispatch of 2 nearby police officers and 1 police car to handle the situation. when When ≤1, it means that a traffic incident has occurred in the j-th monitoring area of ​​the i-th street, and two traffic police officers and one tow truck need to be dispatched to handle it.

[0014] A public security risk early warning system integrating urban road networks and mobile communication networks includes; The data acquisition module is used to deploy a traffic sensing terminal in the j-th monitoring area of ​​the i-th street in the urban road network, and a mobile communication base station covering the j-th monitoring area of ​​the i-th street, to collect data on the frequency of repeated vehicle entry and exit, the number of vehicles staying in prohibited parking locations, the number of vehicles, the speed of pedestrian flow gathering, and the number of people in the j-th monitoring area of ​​the i-th street, respectively. The dataset construction module, based on the mobile communication base station of the j-th monitoring area of ​​the i-th street, collects data on the frequency of repeated vehicle entry and exit, the number of vehicles staying at prohibited parking locations, the number of vehicles, the speed of pedestrian gathering, and the number of people in the j-th monitoring area of ​​the i-th street, to construct a traffic management risk dataset. The model building module is used to construct a spatiotemporal feature fusion model using a convolutional neural network, and inputs the traffic risk dataset into the spatiotemporal feature fusion model to output the prediction result of the j-th monitoring area of ​​the i-th street; The extraction module is used to extract the frequency of repeated vehicle entry and exit, the number of times vehicles stay in prohibited parking locations, the number of vehicles, the speed of pedestrian flow gathering, and the number of people in the j-th monitoring area of ​​the i-th street, based on the traffic risk dataset. The first calculation module calculates the abnormal vehicle behavior coefficient in the j-th monitoring area of ​​the i-th street based on the frequency of repeated vehicle entry and exit, the number of vehicles staying in prohibited parking areas, and the number of vehicles. ; The second calculation module calculates the abnormal behavior coefficient of the population in the j-th monitoring area of ​​the i-th street based on the population size. ; The third calculation module is based on the pedestrian flow aggregation speed and the number of people in the j-th monitoring area of ​​the i-th street. Calculate the population growth index of the j-th monitoring area on the i-th street. ; Combined module, used for the abnormal vehicle behavior coefficients in the j-th monitoring area of ​​the i-th street. Population Abnormal Behavior Coefficient and personnel growth index Combined, construct the ratio coefficient of human and vehicle anomalies. ; The evaluation module compares the ratio of pedestrian and vehicle anomalies with 1. When the ratio is greater than 1, it is classified as a public security incident; when it is less than or equal to 1, it is classified as a traffic incident.

[0015] The traffic risk early warning method and system integrating urban road networks and mobile communication networks provided by this invention have the following beneficial effects: 1. By simultaneously collecting multi-dimensional data such as vehicle entry and exit frequency, number of people staying in no-parking locations, number of people, and speed of pedestrian gathering in the traffic sensing terminals and mobile communication base stations in the monitoring areas of each street in the urban road network, a traffic risk dataset integrating traffic and public security characteristics is constructed, which effectively improves the comprehensive perception capability of urban operation status. By establishing a spatiotemporal feature fusion model through convolutional neural networks, which comprehensively considers the dynamic characteristics of traffic flow and the temporal change pattern of the crowd, it can automatically learn risk change patterns in a multi-dimensional feature space. Compared with the traditional rule threshold judgment method, it has higher risk prediction accuracy and response speed.

[0016] 2. Based on the abnormal behavior coefficient of vehicles, the abnormal behavior coefficient of crowds, and the population growth index, a human-vehicle anomaly ratio coefficient is constructed to realize dynamic correlation analysis between "people-vehicle-area". It can accurately reflect the difference between public security anomalies and traffic anomalies. By comparing the human-vehicle anomaly ratio coefficient with a preset threshold, when the coefficient is greater than 1, it is automatically judged as a public security event, and when it is less than or equal to 1, it is judged as a traffic event. This enables adaptive identification of event types, avoids misjudgment and duplicate scheduling, and significantly improves the joint response efficiency of the public security and traffic police systems. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained from these drawings without creative effort.

[0018] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a system block diagram of the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, 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.

[0020] Example 1, referring to Figure 1 This invention provides a technical solution: a security risk early warning method integrating urban road networks and mobile communication networks, comprising: S1. Based on the traffic sensing terminal deployed in the j-th monitoring area of ​​the i-th street in the urban road network, and the mobile communication base station covering the j-th monitoring area of ​​the i-th street, collect the frequency of repeated vehicle entry and exit, the number of vehicles staying in prohibited parking locations, the number of vehicles, the speed of pedestrian flow gathering, and the number of people in the j-th monitoring area of ​​the i-th street, and construct a traffic management risk dataset. S2. Using a convolutional neural network, construct a spatiotemporal feature fusion model, input the traffic risk dataset into the spatiotemporal feature fusion model, and output the risk prediction results; S3. Based on the traffic risk dataset, construct the abnormal vehicle behavior coefficients for the j-th monitoring area on the i-th street. Population Abnormal Behavior Coefficient and personnel growth index ; S4. Vehicle abnormal behavior coefficients based on the j-th monitoring area of ​​the i-th street. Population Abnormal Behavior Coefficient and personnel growth index Combined, construct the ratio coefficient of human and vehicle anomalies. And preset anomaly thresholds, when the ratio of people and vehicles to anomalies is... When the ratio is greater than 1, it is determined to be a public security incident; when the ratio of the abnormal ratio of people and vehicles is less than or equal to 1, it is determined to be a traffic incident.

[0021] In this embodiment, the present invention, based on traffic sensing terminals deployed in monitoring areas of various streets in the urban road network and mobile communication base stations in the coverage area, synchronously collects multi-dimensional data such as vehicle entry and exit frequency, number of people staying in no-parking locations, number of people, and speed of pedestrian gathering, and constructs a traffic risk dataset that integrates traffic and public security features. This effectively improves the comprehensive perception capability of urban operation status. By establishing a spatiotemporal feature fusion model through convolutional neural networks, it comprehensively considers the dynamic features of traffic flow and the temporal change patterns of the crowd, and can automatically learn risk change patterns in a multi-dimensional feature space. Compared with the traditional rule threshold judgment method, it has higher risk prediction accuracy and response speed.

[0022] Based on the abnormal behavior coefficients of vehicles, abnormal behavior coefficients of crowds, and population growth index, a human-vehicle anomaly ratio coefficient is constructed to realize dynamic correlation analysis between "people-vehicle-area". It can accurately reflect the difference between public security anomalies and traffic anomalies. By comparing the human-vehicle anomaly ratio coefficient with a preset threshold, when the coefficient is greater than 1, it is automatically identified as a public security event, and when it is less than or equal to 1, it is identified as a traffic event. This enables adaptive identification of event types, avoids misjudgment and duplicate scheduling, and significantly improves the joint response efficiency of the public security and traffic police systems.

[0023] Example 2 is an explanation of Example 1; please refer to it. Figure 1 Specifically, S1 includes; S11. By installing a multi-view high-definition camera in the j-th monitoring area on the i-th street, vehicle images in the j-th monitoring area are acquired using the high-definition camera and transmitted to an image analysis terminal via mobile communication. After noise reduction processing, vehicle textures in the images are identified to obtain the number of vehicles in the j-th monitoring area. An image recognition algorithm is then used to identify the license plate of each vehicle in the j-th monitoring area, and the frequency of repeated entry and exit of vehicles in the j-th monitoring area on the i-th street is monitored. ; S12. Take images of the no-parking locations in the j-th monitoring area using a high-definition camera, and transmit them to an image analysis terminal via mobile communication. Identify vehicle texture features in the images and obtain the number of vehicles staying at the no-parking locations in the j-th monitoring area on the i-th street. .

[0024] In this embodiment, vehicle images within the monitoring area are acquired using a multi-view high-definition camera, and denoising and texture recognition processing are performed on the image analysis terminal. This enables accurate extraction of vehicle outlines, body textures, and license plate features under different lighting and occlusion conditions, effectively improving the accuracy of vehicle detection and recognition. By using the license plate recognition results to statistically analyze the multiple appearances of the same vehicle in the monitoring area, the frequency of repeated vehicle entry and exit can be automatically calculated, reflecting the level of traffic activity and abnormal traffic behavior in the area, making it easier to detect suspicious vehicles patrolling abnormally or vehicles posing a security risk.

[0025] By continuously monitoring no-parking areas in the monitoring zone using high-definition cameras and combining image recognition algorithms to automatically determine the status of vehicles in these areas and obtain the number of vehicles parked there, the system can quickly identify and locate vehicles that have been lingering or illegally parked for extended periods. This provides accurate data support for traffic management and security control. The system also utilizes mobile communication networks to transmit image data to image analysis terminals in real time, without relying on local storage or manual collection. This significantly improves the timeliness of data processing and the system's response speed, ensuring the continuity of abnormal vehicle behavior detection and real-time early warning capabilities.

[0026] Example 3 is an explanation of Example 1; please refer to the provided text. Figure 1 Specifically, S1 also includes; S13. Based on the high-definition camera covering the j-th monitoring area of ​​the i-th street, capture images of the j-th monitoring area in real time, and transmit them to the image analysis terminal via mobile communication. Enhance the images, use a human target detection algorithm to identify the number of people in the images, and obtain the number of people in the j-th monitoring area of ​​the i-th street. And collect the number of people in the j-th monitoring area of ​​the i-th street per unit time. Calculate the pedestrian flow aggregation speed in the j-th monitoring area of ​​the i-th street. This is used to construct a traffic risk dataset.

[0027] In this embodiment, by deploying high-definition cameras to collect images of the monitoring area in real time, and using a human target detection algorithm in the image analysis terminal, the number of individuals in the image can be accurately identified under different density, lighting and viewing angle conditions, and the number of people in the j-th monitoring area of ​​the i-th street can be obtained. This effectively avoids the errors caused by traditional manual statistics or low-precision sensors. By calculating the speed of crowd gathering by the change in the number of people within a continuous time window, the system can reflect the real-time trend of crowd growth or evacuation, providing dynamic evidence for identifying abnormal gatherings and precursors of emergencies, and improving the system's sensitivity to security risks.

[0028] Example 4 is an explanation of Example 1; please refer to the provided text. Figure 1 Specifically, S2 includes; S21. Construct a spatiotemporal feature fusion model using a convolutional neural network, train and test the spatiotemporal feature fusion model with a traffic risk dataset, and use the trained spatiotemporal feature fusion model as a traffic risk test and evaluation model. At the same time, use the intermediate layer output of the device running spatiotemporal feature fusion model as a feature vector to identify feature information, and use the trained spatiotemporal feature fusion model as a data running prediction.

[0029] In this embodiment, a convolutional neural network structure is adopted, which can simultaneously extract vehicle behavior features, crowd behavior features, and time series change patterns from traffic risk datasets, thereby achieving joint modeling of spatial distribution features and temporal dynamic features and effectively improving the model's ability to represent complex urban dynamic scenarios.

[0030] Example 5 is an explanation of Example 1; please refer to it. Figure 1 Specifically, S3 includes; S31. Frequency of repeated vehicle entry and exit in the j-th monitoring area of ​​the i-th street based on the traffic risk dataset. Number of vehicles staying in no-parking zones Number of vehicles The abnormal vehicle behavior coefficient of the j-th monitoring area on the i-th street is calculated using the following method. ; First, the frequency of vehicle entry and exit in the j-th monitoring area of ​​the i-th street is repeatedly counted. Number of vehicles staying in no-parking zones Number of vehicles After normalization, the frequency of repeated vehicle entry and exit in the j-th monitoring area of ​​the i-th street is obtained. Number of vehicles staying in no-parking zones Number of vehicles ; In the formula, This represents the maximum frequency of repeated vehicle entry and exit, which can be directly obtained based on historical data. In the formula, This represents the maximum number of vehicles that can remain in prohibited parking areas, which can be directly obtained based on historical data. In the formula, This represents the maximum number of vehicles, which can be directly obtained based on historical data. Secondly, based on the normalized frequency of repeated vehicle entry and exit in the j-th monitoring area of ​​the i-th street. Number of vehicles staying in no-parking zones Number of vehicles The abnormal vehicle behavior coefficient of the j-th monitoring area on the i-th street is calculated using the following formula. ; ; In the formula , and This represents the weighting coefficient, based on the frequency of repeated entry and exit of vehicles in the j-th monitoring area of ​​the i-th street in historical data. Number of vehicles staying in no-parking zones Number of vehicles The abnormal vehicle behavior coefficient of the j-th monitoring area on the i-th street The degree of influence is determined by the percentage of influence, and combined with the historical weighting coefficients, to... , and The value is adjusted.

[0031] In this embodiment, by normalizing multi-dimensional features such as the frequency of repeated vehicle entry and exit, the number of times vehicles stop at prohibited parking locations, and the number of vehicles, the differences between different feature dimensions are effectively eliminated, allowing each feature to participate in the calculation on the same scale. This achieves a unified quantitative expression of abnormal vehicle behavior. By using weighted coefficients, differentiated weights are assigned according to the degree of influence of different features on abnormal behavior, which can flexibly adjust the model sensitivity and improve the system's ability and stability to identify different types of traffic anomalies (such as frequent passage, illegal parking, and vehicle congestion).

[0032] By calculating the abnormal vehicle behavior coefficient, the system maps vehicle behavior from raw monitoring data to quantitative indicators of abnormal risk, enabling it to assess the degree of vehicle abnormality in different monitoring areas in a numerical form, and providing a reliable quantitative basis for subsequent classification of public security and traffic incidents.

[0033] Example 6 is an explanation of Example 1; please refer to the provided text. Figure 1 Specifically, S3 also includes; S32. By presetting the abnormal vehicle behavior threshold A; The average and standard deviation of the abnormal vehicle behavior coefficient are calculated based on historical street data. The sum of the average and standard deviation is used as the abnormal vehicle behavior threshold A. The abnormal vehicle behavior coefficient of the j-th monitoring area on the i-th street Compare with the vehicle abnormal behavior threshold A to generate a vehicle abnormal behavior assessment instruction, including: when When >A, it indicates that the vehicle behavior in the j-th monitoring area of ​​the i-th street is abnormal, and a vehicle abnormality alarm instruction is sent to the traffic management terminal to dispatch nearby traffic police to handle the situation and direct traffic. when When A ≤ A, it means that the vehicle behavior in the j-th monitoring area of ​​the i-th street is normal.

[0034] In this embodiment, by setting a threshold for abnormal vehicle behavior, the calculated abnormal vehicle behavior coefficient is judged in real time, realizing the integrated linkage from "abnormal detection" to "early warning response". This enables abnormal traffic events to be identified in a timely manner and automatically generate handling instructions, improving the automation level of urban traffic management. By using the quantitative threshold A to judge abnormal behavior, alarms can be quickly triggered when the frequency of repeated vehicle entry and exit, the number of illegal parking, or the vehicle density are significantly abnormal, effectively avoiding traffic congestion or safety hazards caused by delays in manual judgment, and improving the accuracy and real-time performance of traffic event early warning.

[0035] Example 7 is an explanation of Example 1; please refer to it. Figure 1 Specifically, S3 also includes; S33. Population size in the j-th monitoring area of ​​the i-th street based on the traffic risk dataset. The abnormal behavior coefficients of the population in the j-th monitoring area of ​​the i-th street are obtained through the following method. ; In the formula Expressed as a unit of time, It represents the historical average number of people in the j-th monitoring area of ​​the i-th street within a unit of time. It is expressed as the historical standard deviation of the population size in the j-th monitoring area of ​​the i-th street within a unit of time. S34. By setting a threshold S for abnormal crowd behavior, the coefficient of abnormal crowd behavior in the j-th monitoring area of ​​the i-th street is... Compare with the abnormal behavior threshold S of the population to generate an abnormal behavior assessment instruction for the population, including: when When the value is greater than S, it indicates that the number of people gathering in the j-th monitoring area of ​​the i-th street is abnormal, and an abnormal alarm instruction is sent to the public security command terminal, requiring nearby police officers to be dispatched to evacuate the crowd. when When ≤S, it means that the number of people in the j-th monitoring area of ​​the i-th street is normal.

[0036] In this embodiment, by calculating the standardized deviation of the number of people per unit time from the historical average and standard deviation, the degree of abnormality in the fluctuation of the number of people in the monitoring area is effectively reflected. This allows for the rapid identification of abnormal crowd gatherings such as parties, assemblies, and sudden crowd surges, significantly improving the early detection capability of security risks. When the signal is greater than 5 seconds, the system automatically identifies an abnormal crowd gathering and sends an alarm to the public security command terminal without manual intervention. This enables real-time detection and automated early warning of security incidents, effectively shortening the response time.

[0037] After identifying abnormal gatherings, the system automatically dispatches nearby police officers to the scene for evacuation. Through algorithm-driven police force allocation, it improves resource utilization efficiency, realizes closed-loop management from data analysis to emergency response, and enhances the level of public security and prevention during high-risk periods such as nighttime and holidays.

[0038] Example 8 is an explanation of Example 1; please refer to it. Figure 1 Specifically, S3 also includes; S35. Population count in the j-th monitoring area of ​​the i-th street. The pedestrian flow aggregation speed in the j-th monitoring area of ​​the i-th street is calculated using the following formula. ; In the formula, The time interval is represented by the number of people collected. Subsequently, based on the pedestrian flow aggregation speed in the j-th monitoring area of ​​the i-th street... The population growth index of the j-th monitoring area on the i-th street is calculated using the following formula. ; In the formula, This is expressed as a time span factor, which is determined by selecting an appropriate multiple based on the sampling frequency setting and the interval between continuous data collections per unit time. .

[0039] In this embodiment, the rate of population gathering is calculated by the rate of change in the number of people per unit time, which can accurately depict the growth or dispersal trend of the number of people in the monitoring area. It expands from static population density monitoring to dynamic change monitoring, improves the system's sensitivity to sudden gathering events, and uses the population growth index to index the population change, which can reflect the acceleration characteristics of population growth. When the growth index shows a rapid upward trend, it can predict potential gathering risks in advance, which helps to realize the transformation of security incidents from "post-event response" to "pre-event warning".

[0040] By introducing a time span factor, the time window can be flexibly adjusted according to different blocks or time periods (such as holidays, night markets, etc.), enabling dual identification of short-term anomalies (such as temporary gatherings) and long-term trends (such as continuous gatherings), improving the model's spatiotemporal adaptability. Combining the speed of crowd gathering with the population growth index can simultaneously capture both the "rate" and "trend" characteristics of changes in the number of people, helping to distinguish between normal mobile gatherings and abnormal stagnant gatherings, thereby reducing misjudgments and improving the accuracy of security risk classification.

[0041] Example 9, this example is an explanation of Example 1, please refer to it. Figure 1 Specifically, S4 includes: S41. Based on the abnormal vehicle behavior coefficient of the j-th monitoring area on the i-th street. Population Abnormal Behavior Coefficient and personnel growth index Combined, the ratio coefficient of pedestrian and vehicle anomalies in the j-th monitoring area of ​​the i-th street is calculated using the following formula. ; First, calculate the abnormal vehicle behavior coefficient for the j-th monitoring area on the i-th street. By performing quantile mapping according to the historical sample distribution, standardized parameters of vehicle anomalies are obtained. ; The abnormal behavior coefficient of the population in the j-th monitoring area of ​​the i-th street. Mapped to standardized parameters ; The population growth index of the j-th monitoring area on the i-th street The normalized growth index of personnel was obtained by comparing with the historical 95th percentile. ; Then, the pedestrian-vehicle anomaly ratio coefficient for the j-th monitoring area on the i-th street is calculated using the following formula. ; In the formula Represented as a natural number divisible by zero, with a value of 0.001; S42. Calculate the ratio of pedestrian and vehicle anomalies in the j-th monitoring area of ​​the i-th street. Compare with 1; when When the value is greater than 1, it indicates that a public security incident has occurred in the j-th monitoring area of ​​the i-th street, requiring the dispatch of 2 nearby police officers and 1 police car to handle the situation. when When ≤1, it means that a traffic incident has occurred in the j-th monitoring area of ​​the i-th street, and two traffic police officers and one tow truck need to be dispatched to handle it.

[0042] In this embodiment, by integrating the abnormal behavior characteristics of the crowd and vehicles, and using the fusion calculation of the abnormal behavior coefficient of the crowd, the abnormal behavior coefficient of the vehicles, and the growth index of the population, a human-vehicle abnormality ratio coefficient is constructed. When the ratio coefficient exceeds a set threshold, it is automatically determined to be a public security type event; otherwise, it is a traffic type event. This realizes the intelligent differentiation between traffic and public security risks, avoiding the subjectivity and delay of manual judgment.

[0043] The ratio of human-vehicle anomalies is clearly defined by a mathematical formula, and its numerical change can intuitively reflect the dominant relationship between human and vehicle dynamics. When human anomalies dominate ( >1) is a security risk, when the vehicle is abnormally dominant ( When the threshold is ≤1), it represents a traffic risk, providing quantifiable and traceable decision-making basis and enhancing the interpretability of the model results. Based on the judgment result, the system automatically sends response instructions to the public security or traffic management dispatch system. When it is judged as a public security incident, it automatically dispatches police officers and police cars to the scene. When it is judged as a traffic incident, it automatically dispatches traffic police and tow trucks to handle the situation. This realizes cross-departmental collaboration and intelligent resource allocation, and greatly improves the response speed and handling efficiency of emergencies.

[0044] Example 10: This example is an explanation of Example 1. Please refer to the provided text. Figure 2 Specifically, a security risk early warning system integrating urban road networks and mobile communication networks includes; The data acquisition module is used to deploy a traffic sensing terminal in the j-th monitoring area of ​​the i-th street in the urban road network, and a mobile communication base station covering the j-th monitoring area of ​​the i-th street, to collect data on the frequency of repeated vehicle entry and exit, the number of vehicles staying in prohibited parking locations, the number of vehicles, the speed of pedestrian flow gathering, and the number of people in the j-th monitoring area of ​​the i-th street, respectively. The dataset construction module, based on the mobile communication base station of the j-th monitoring area of ​​the i-th street, collects data on the frequency of repeated vehicle entry and exit, the number of vehicles staying at prohibited parking locations, the number of vehicles, the speed of pedestrian gathering, and the number of people in the j-th monitoring area of ​​the i-th street, to construct a traffic management risk dataset. The model building module is used to construct a spatiotemporal feature fusion model using a convolutional neural network, and inputs the traffic risk dataset into the spatiotemporal feature fusion model to output the prediction result of the j-th monitoring area of ​​the i-th street; The extraction module is used to extract the frequency of repeated vehicle entry and exit, the number of times vehicles stay in prohibited parking locations, the number of vehicles, the speed of pedestrian flow gathering, and the number of people in the j-th monitoring area of ​​the i-th street, based on the traffic risk dataset. The first calculation module calculates the abnormal vehicle behavior coefficient in the j-th monitoring area of ​​the i-th street based on the frequency of repeated vehicle entry and exit, the number of vehicles staying in prohibited parking areas, and the number of vehicles. ; The second calculation module calculates the abnormal behavior coefficient of the population in the j-th monitoring area of ​​the i-th street based on the population size. ; The third calculation module is based on the pedestrian flow aggregation speed and the number of people in the j-th monitoring area of ​​the i-th street. Calculate the population growth index of the j-th monitoring area on the i-th street. ; Combined module, used for the abnormal vehicle behavior coefficients in the j-th monitoring area of ​​the i-th street. Population Abnormal Behavior Coefficient and personnel growth index Combined, construct the ratio coefficient of human and vehicle anomalies. ; The evaluation module compares the ratio of pedestrian and vehicle anomalies with 1. When the ratio is greater than 1, it is classified as a public security incident; when it is less than or equal to 1, it is classified as a traffic incident.

[0045] In this embodiment, the system integrates traffic perception data from the urban road network with crowd dynamic information from the mobile communication network. It can simultaneously perceive changes in vehicle behavior and crowd gathering, construct a unified traffic management risk dataset, and realize integrated monitoring and early warning of traffic incidents and public security incidents. This improves the comprehensiveness and accuracy of urban public safety incident identification. The data acquisition module can simultaneously access traffic perception terminals and communication base stations to collect multi-dimensional features in real time, such as the frequency of repeated vehicle entry and exit, the number of prohibited stops, the number of people, and the gathering speed. This achieves multi-source heterogeneous fusion of urban dynamic elements and effectively overcomes the problem of low identification accuracy of single data sources in complex scenarios.

[0046] The model building module utilizes convolutional neural networks to establish a spatiotemporal feature fusion model, jointly modeling the spatial correlation and temporal evolution patterns in the traffic risk dataset. This improves the accuracy and generalization ability of event risk prediction, enabling early identification of abnormal trends and early warning of risks. The system is configured with first, second, and third calculation modules to calculate vehicle abnormal behavior coefficients, crowd abnormal behavior coefficients, and population growth index, respectively. Through quantitative modeling of the dynamic characteristics of traffic and crowds, it achieves a refined characterization of behaviors such as traffic congestion, illegal parking, and crowd gathering, helping to discover potential abnormal patterns and risk factors. The combined module comprehensively calculates vehicle and crowd behavior characteristics to construct the human-vehicle abnormal ratio coefficient Kpc,ij. The evaluation module automatically determines the event type by comparing it with a threshold of 1. >1 is considered a public security incident, requiring the dispatch of police officers and police vehicles; when When the value is ≤1, it is a traffic-related incident, requiring the dispatch of traffic police and tow trucks to manage the situation. This system enables automated identification of incident types and intelligent coordination of emergency resources.

[0047] Example; Vehicle abnormal behavior coefficient in the j-th monitoring area of ​​the i-th street =0.4, and quantile mapping is performed according to the historical sample distribution to obtain the standardized parameters of vehicle anomalies. =0.4; The population growth index of the j-th monitoring area on the i-th street =0.6, and the normalized velocity parameter was obtained by comparing it with the historical 95th percentile. =0.6; The abnormal behavior coefficient of the population in the j-th monitoring area of ​​the i-th street. =0.8, mapped to standardized parameters =0.8; ; 3.19 > 1, indicating a public security incident has occurred in the j-th monitoring area of ​​the i-th street, requiring the dispatch of 2 nearby police officers and 1 police car to handle the situation.

[0048] The threshold is set to facilitate comparison. The size of the threshold depends on the amount of sample data and the number of bases set by those skilled in the art for each set of sample data; as long as it does not affect the ratio between the parameter and the quantized value, it is acceptable.

[0049] The above formulas are all derived from software simulation using a large amount of data, and are selected to be close to the actual values. The coefficients in the formulas are set by those skilled in the art according to the actual situation. The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or changes made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A traffic risk early warning method integrating urban road networks and mobile communication networks, characterized in that, include: S1. Based on the traffic sensing terminal deployed in the j-th monitoring area of ​​the i-th street in the urban road network, and the mobile communication base station covering the j-th monitoring area of ​​the i-th street, collect the frequency of repeated vehicle entry and exit, the number of vehicles staying in prohibited parking locations, the number of vehicles, the speed of pedestrian flow gathering, and the number of people in the j-th monitoring area of ​​the i-th street, and construct a traffic management risk dataset. S2. Using a convolutional neural network, construct a spatiotemporal feature fusion model, input the traffic risk dataset into the spatiotemporal feature fusion model, and output the risk prediction results; S3. Based on the traffic risk dataset, construct the abnormal vehicle behavior coefficients for the j-th monitoring area on the i-th street. Population Abnormal Behavior Coefficient and personnel growth index ; S4. Vehicle abnormal behavior coefficients based on the j-th monitoring area of ​​the i-th street. Population Abnormal Behavior Coefficient and personnel growth index Combined, construct the ratio coefficient of human and vehicle anomalies. And preset anomaly thresholds, when the ratio of people and vehicles to anomalies is... When the ratio is greater than 1, it is determined to be a public security incident; when the ratio of the abnormal ratio of people and vehicles is less than or equal to 1, it is determined to be a traffic incident.

2. The traffic risk early warning method integrating urban road network and mobile communication network according to claim 1, characterized in that, S1 includes; S11. By installing a multi-view high-definition camera in the j-th monitoring area on the i-th street, vehicle images in the j-th monitoring area are acquired using the high-definition camera and transmitted to an image analysis terminal via mobile communication. After noise reduction processing, vehicle textures in the images are identified to obtain the number of vehicles in the j-th monitoring area. An image recognition algorithm is then used to identify the license plate of each vehicle in the j-th monitoring area, and the frequency of repeated entry and exit of vehicles in the j-th monitoring area on the i-th street is monitored. ; S12. Take images of the no-parking locations in the j-th monitoring area using a high-definition camera, and transmit them to an image analysis terminal via mobile communication. Identify vehicle texture features in the images and obtain the number of vehicles staying at the no-parking locations in the j-th monitoring area on the i-th street. .

3. The traffic risk early warning method integrating urban road network and mobile communication network according to claim 2, characterized in that, S1 also includes; S13. Based on the high-definition camera covering the j-th monitoring area of ​​the i-th street, capture images of the j-th monitoring area in real time, and transmit them to the image analysis terminal via mobile communication. Enhance the images, use a human target detection algorithm to identify the number of people in the images, and obtain the number of people in the j-th monitoring area of ​​the i-th street. And collect the number of people in the j-th monitoring area of ​​the i-th street per unit time. Calculate the pedestrian flow aggregation speed in the j-th monitoring area of ​​the i-th street. This is used to construct a traffic risk dataset.

4. The traffic risk early warning method integrating urban road network and mobile communication network according to claim 3, characterized in that, S2 includes; S21. Construct a spatiotemporal feature fusion model using a convolutional neural network, train and test the spatiotemporal feature fusion model with a traffic risk dataset, and use the trained spatiotemporal feature fusion model as a traffic risk test and evaluation model. At the same time, use the intermediate layer output of the device running spatiotemporal feature fusion model as a feature vector to identify feature information, and use the trained spatiotemporal feature fusion model as a data running prediction.

5. The traffic risk early warning method integrating urban road network and mobile communication network according to claim 4, characterized in that, S3 includes; S31. Frequency of repeated vehicle entry and exit in the j-th monitoring area of ​​the i-th street based on the traffic risk dataset. Number of vehicles staying in no-parking zones Number of vehicles The abnormal vehicle behavior coefficient of the j-th monitoring area on the i-th street is calculated using the following method. ; First, the frequency of vehicle entry and exit in the j-th monitoring area of ​​the i-th street is repeatedly counted. Number of vehicles staying in no-parking zones Number of vehicles After normalization, the frequency of repeated vehicle entry and exit in the j-th monitoring area of ​​the i-th street is obtained. Number of vehicles staying in no-parking zones Number of vehicles ; In the formula, This represents the maximum frequency of repeated vehicle entry and exit. In the formula, This represents the maximum number of vehicles that can remain in a no-parking zone. In the formula, This represents the maximum number of vehicles. Secondly, based on the normalized frequency of repeated vehicle entry and exit in the j-th monitoring area of ​​the i-th street. Number of vehicles staying in no-parking zones Number of vehicles The abnormal vehicle behavior coefficient of the j-th monitoring area on the i-th street is calculated using the following formula. ; ； In the formula , and This represents the weighting coefficient.

6. The traffic risk early warning method integrating urban road network and mobile communication network according to claim 5, characterized in that, S3 also includes; S32. By setting a preset abnormal vehicle behavior threshold A, the abnormal vehicle behavior coefficient of the j-th monitoring area on the i-th street is... Compare with the vehicle abnormal behavior threshold A to generate a vehicle abnormal behavior assessment instruction, including: when When >A, it indicates that the vehicle behavior in the j-th monitoring area of ​​the i-th street is abnormal, and a vehicle abnormality alarm instruction is sent to the traffic management terminal to dispatch nearby traffic police to handle the situation and direct traffic. when When A ≤ A, it means that the vehicle behavior in the j-th monitoring area of ​​the i-th street is normal.

7. The traffic risk early warning method integrating urban road network and mobile communication network according to claim 6, characterized in that, S3 also includes; S33. Population size in the j-th monitoring area of ​​the i-th street based on the traffic risk dataset. The abnormal behavior coefficients of the population in the j-th monitoring area of ​​the i-th street are obtained through the following method. ; In the formula Expressed as a unit of time, It represents the historical average number of people in the j-th monitoring area of ​​the i-th street within a unit of time. It is expressed as the historical standard deviation of the population size in the j-th monitoring area of ​​the i-th street within a unit of time. S34. By setting a threshold S for abnormal crowd behavior, the coefficient of abnormal crowd behavior in the j-th monitoring area of ​​the i-th street is... Compare with the abnormal behavior threshold S of the population to generate an abnormal behavior assessment instruction for the population, including: when When the value is greater than S, it indicates that the number of people gathering in the j-th monitoring area of ​​the i-th street is abnormal, and an abnormal alarm instruction is sent to the public security command terminal, requiring nearby police officers to be dispatched to evacuate the crowd. when When ≤S, it means that the number of people in the j-th monitoring area of ​​the i-th street is normal.

8. The traffic risk early warning method integrating urban road network and mobile communication network according to claim 7, characterized in that, S3 also includes; S35, Based on the pedestrian flow aggregation speed in the j-th monitoring area of ​​the i-th street. The population growth index of the j-th monitoring area on the i-th street is calculated using the following formula. ; In the formula, It is represented as a time span factor.

9. The traffic risk early warning method integrating urban road network and mobile communication network according to claim 8, characterized in that, S4 includes; S41. Based on the abnormal vehicle behavior coefficient of the j-th monitoring area on the i-th street. Population Abnormal Behavior Coefficient and personnel growth index Combined, the ratio coefficient of pedestrian and vehicle anomalies in the j-th monitoring area of ​​the i-th street is calculated using the following formula. ; First, calculate the abnormal vehicle behavior coefficient for the j-th monitoring area on the i-th street. By performing quantile mapping according to the historical sample distribution, standardized parameters of vehicle anomalies are obtained. ; The abnormal behavior coefficient of the population in the j-th monitoring area of ​​the i-th street. Mapped to standardized parameters ; The population growth index of the j-th monitoring area on the i-th street The normalized growth index of personnel was obtained by comparing with the historical 95th percentile. ; Then, the pedestrian-vehicle anomaly ratio coefficient for the j-th monitoring area on the i-th street is calculated using the following formula. ; In the formula Represented as a natural number divisible by zero, with a value of 0.001; S42. Calculate the ratio of pedestrian and vehicle anomalies in the j-th monitoring area of ​​the i-th street. Compare with 1; when When the value is greater than 1, it indicates that a public security incident has occurred in the j-th monitoring area of ​​the i-th street, requiring the dispatch of 2 nearby police officers and 1 police car to handle the situation. when When ≤1, it means that a traffic incident has occurred in the j-th monitoring area of ​​the i-th street, and two traffic police officers and one tow truck need to be dispatched to handle it.

10. A traffic risk early warning system integrating urban road networks and mobile communication networks, applied to the traffic risk early warning method integrating urban road networks and mobile communication networks as described in claims 1-9, characterized in that, include; The data acquisition module is used to deploy a traffic sensing terminal in the j-th monitoring area of ​​the i-th street in the urban road network, and a mobile communication base station covering the j-th monitoring area of ​​the i-th street, to collect data on the frequency of repeated vehicle entry and exit, the number of vehicles staying in prohibited parking locations, the number of vehicles, the speed of pedestrian flow gathering, and the number of people in the j-th monitoring area of ​​the i-th street, respectively. The dataset construction module, based on the mobile communication base station of the j-th monitoring area of ​​the i-th street, collects data on the frequency of repeated vehicle entry and exit, the number of vehicles staying at prohibited parking locations, the number of vehicles, the speed of pedestrian gathering, and the number of people in the j-th monitoring area of ​​the i-th street, to construct a traffic management risk dataset. The model building module is used to construct a spatiotemporal feature fusion model using a convolutional neural network, and inputs the traffic risk dataset into the spatiotemporal feature fusion model to output the prediction result of the j-th monitoring area of ​​the i-th street; The extraction module is used to extract the frequency of repeated vehicle entry and exit, the number of times vehicles stay in prohibited parking locations, the number of vehicles, the speed of pedestrian flow gathering, and the number of people in the j-th monitoring area of ​​the i-th street, based on the traffic risk dataset. The first calculation module calculates the abnormal vehicle behavior coefficient in the j-th monitoring area of ​​the i-th street based on the frequency of repeated vehicle entry and exit, the number of vehicles staying in prohibited parking areas, and the number of vehicles. ; The second calculation module calculates the abnormal behavior coefficient of the population in the j-th monitoring area of ​​the i-th street based on the population size. ; The third calculation module is based on the pedestrian flow aggregation speed and the number of people in the j-th monitoring area of ​​the i-th street. Calculate the population growth index of the j-th monitoring area on the i-th street. ; Combined module, used for the abnormal vehicle behavior coefficients in the j-th monitoring area of ​​the i-th street. Population Abnormal Behavior Coefficient and personnel growth index Combined, construct the ratio coefficient of human and vehicle anomalies. ; The evaluation module compares the ratio of pedestrian and vehicle anomalies with 1. When the ratio is greater than 1, it is classified as a public security incident; when it is less than or equal to 1, it is classified as a traffic incident.

Images