Safety early warning signal sending control method applied to intelligent monitoring equipment

Abstract

The invention discloses a safety early warning signal sending control method applied to intelligent monitoring equipment, and belongs to the technical field of artificial intelligence. According to the invention, when a traffic accident occurs, monitoring data, meteorological data and geographic data of a place where the traffic accident occurs are acquired to form traffic accident data; extracting visual features, weather features and road environment features; identifying whether the vehicle violates a rule specified by the traffic sign, judging whether the driving track of the vehicle accords with a lane rule, and forming a traffic violation risk index; combining weather characteristics and road environment characteristics to form an environment interference intensity index; checking the damage condition of traffic signs around the accident site and the shielding condition of buildings to pedestrians to obtain a road facility association index; training the multi-layer perceptron model, inputting index data, and outputting actual proportions of human factors and natural factors of traffic accidents; and when the proportion of the natural factors exceeds a threshold value, providing a reconstruction suggestion of a traffic accident occurrence place.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a safety warning signal sending control method applied to intelligent monitoring equipment. Background Art

[0002] With the acceleration of urbanization and the continuous increase in the number of motor vehicles, frequent traffic accidents have become a serious social problem, causing a huge impact on people's lives and property safety and social and economic development. Accurately and quickly analyzing the causes of traffic accidents is crucial for efficiently handling accidents and preventing similar accidents from happening again.

[0003] Previous methods were not sophisticated enough in extracting target features from surveillance videos and could only identify the rough outlines of vehicles and pedestrians. It was difficult to accurately capture key features such as the instantaneous turning angle of vehicles, changes in pedestrian speed, and detailed status of traffic signs. Risk assessment models constructed using existing technologies often focus on a single factor, either focusing on abnormal vehicle driving trajectories or simply considering the severity of the weather, without comprehensively considering the synergistic effects of multiple factors such as traffic violations, environmental interference, and road facility conditions. Existing technologies mainly focus on post-accident handling and lack the ability to proactively prevent accidents. It is impossible to predict the occurrence of accidents in advance and take corresponding preventive measures through in-depth analysis of historical accident data and monitoring of potential risk factors. Summary of the invention

[0004] The purpose of the present invention is to provide a safety warning signal sending control method applied to intelligent monitoring equipment to solve the problems raised in the prior art.

[0005] To achieve the above object, the present invention provides the following technical solutions: A safety warning signal sending control method applied to an intelligent monitoring device, the method comprising the following steps: When a traffic accident occurs, obtain monitoring data, meteorological data and geographic data of the accident location, perform data preprocessing and integration, and form traffic accident data; Based on the traffic accident data, the YOLOv5 model is used to extract visual features of targets from the monitoring data, and the targets include vehicles, pedestrians and traffic signs; weather features are extracted based on meteorological data, and precipitation, wind speed, temperature and visibility are quantified as characteristic values ​​of the degree of impact on traffic; road environment features are extracted from geographic data; Based on traffic signs, identify whether the vehicle violates the rules specified by the traffic signs, determine whether the vehicle's driving trajectory complies with the lane rules, and obtain the driver's operation score in combination with weather characteristics to form a traffic violation risk index; combine weather characteristics and road environment characteristics to form an environmental interference intensity index; check the damage of traffic signs around the accident site, obtain the location of pedestrians and the distribution of surrounding buildings, check the obstruction of pedestrians by buildings, and obtain the road facility association index; Collect past traffic accident data, build a multi-layer perceptron model architecture, use the training set data to train the multi-layer perceptron model, continuously adjust the model's weights and biases through the back propagation algorithm, use the traffic violation risk index, environmental interference intensity index, and road facility association index as input, and use the actual proportion of human and natural factors in traffic accidents as the target output; A threshold for the proportion of natural factors is set. When the proportion of natural factors exceeds the threshold, renovation suggestions for the location where the traffic accident occurred are provided based on weather characteristics and road environment characteristics.

[0006] The YOLOv5 model is pre-trained on traffic scene image data. When a traffic accident occurs, the monitoring data in the traffic accident data is input into the YOLOv5 model frame by frame in an image sequence of one frame per second. For vehicles, the model, body color and license plate number are identified, and the driving trajectory of the vehicle before the accident is tracked. For pedestrians, the relative relationship between the pedestrian’s position and the vehicle at the time of the accident is determined. For traffic signs, the sign type and its position in the image are identified.

[0007] Furthermore, based on the weather characteristics, the precipitation is divided into different magnitudes according to meteorological standards, and the depth of water accumulation is estimated in combination with the road environment characteristics. The degree of line of sight obstruction is determined by observing the operating frequency of the wipers, the density of raindrops in front of the camera lens, and the driver's head movement through monitoring data; based on the precipitation magnitude, water accumulation depth, and line of sight obstruction, the precipitation characteristics are quantified; According to the wind speed value in the meteorological data, the corresponding wind force level standard is used. For different wind force level standards and different vehicle models, the impact of wind force on vehicle driving stability is judged and the wind speed characteristics are quantified; Quantify temperature characteristics based on their impact on vehicle components and their association with road conditions; The visibility is classified into different ranges to obtain the visibility range. The driver's operation under different visibility conditions is observed through monitoring data. The influence of the visibility range on the driver's operation is quantified to obtain the visibility characteristics. Furthermore, the quantified precipitation characteristics, wind speed characteristics, temperature characteristics and visibility characteristics are used as weather characteristics.

[0008] Using the geographic information system, the radius of curvature of the curve at the accident site was measured, and combined with the terrain information in the geographic data, it was determined whether there was a slope change at the curve. Based on satellite remote sensing images and field 3D modeling data, the obstruction of the view by vegetation, buildings and terrain undulations around the curve was analyzed. Based on the driver's perspective, the visible range at different positions on the curve was simulated, and the proportion of the obstructed view to the total view was calculated to obtain the curve characteristics. Further, the climbing and downhill sections of the road are identified, the starting and ending points of the climbing or downhill sections are determined according to the altitude change trend in the geographic data, the length and average slope of the climbing and downhill sections are calculated, and the slope characteristics are obtained; Measure the minimum distance between the buildings around the accident site and the road edge, and extract the distribution characteristics of the surrounding buildings; Furthermore, the location of the accident is identified based on the geographic data as to whether it is at an intersection. If so, the type of intersection is further determined, and the traffic lights, signs and markings at the intersection are checked to extract the characteristics of the road intersection. Furthermore, curve characteristics, slope characteristics, surrounding building distribution characteristics and road intersection characteristics are used as road environment characteristics.

[0009] Identify and locate traffic signs, and based on traffic rules, determine vehicle violations and deduct points for violations; use the Hough transform algorithm to identify the position and shape of lane lines, and obtain lane line information by fusing and optimizing the lane line detection results of multiple consecutive frames of images; track the vehicle's driving trajectory on the road based on the lane line information, use the target tracking algorithm to record the vehicle's center position, driving direction, and relative position relationship with the lane line, and determine the vehicle's violations based on traffic rules and deduct points for violations; Furthermore, when the meteorological data shows that the precipitation intensity reaches moderate rain or above, the driver's use of wipers is observed through monitoring data to determine whether the rain has obstructed the driver's vision, and points are deducted for violations; the driving stability of the vehicle is observed, and if the vehicle slips due to accumulated water on the road and the driver does not take measures to slow down or avoid it, points are deducted for violations; operational considerations are made in weather with insufficient visibility, and driving behavior analysis is conducted in windy weather, and points are deducted for violations; Furthermore, the violation points are accumulated to obtain the total violation risk score, and a linear change is adopted to obtain the traffic violation risk index.

[0010] According to the precipitation and precipitation intensity information in the meteorological data, the degree of interference to traffic is quantified; according to the wind speed value in the meteorological data, the interference intensity is judged according to the wind force level standard; according to the temperature in the meteorological data, the degree of impact on automobile parts is judged; and the visibility is graded and quantified according to different ranges; Quantify interference based on road environment characteristics and evaluate the interference values ​​of curve characteristics, slope characteristics, surrounding building distribution characteristics and road intersection characteristics; Furthermore, various interference values ​​based on weather characteristics and various interference values ​​based on road environment characteristics are accumulated to obtain a total environmental interference intensity score; and a linear change is adopted to obtain an environmental interference intensity index.

[0011] For each identified traffic sign, image feature extraction is used to determine whether its surface is damaged, faded, or deformed, and the damage score of the traffic sign is obtained. The source of obstructions around the traffic sign is determined by combining geographic information system data and monitoring images, and the obstruction score of the traffic sign is quantified by measuring the proportion of the obstructions in the field of view directly in front of the traffic sign. The damage score and the obstruction score of the traffic sign are added together to determine the violation points for the traffic sign factor.

[0012] Based on monitoring data, the target tracking algorithm is used to lock pedestrians at the accident scene and surrounding areas in real time, match the pedestrian's position with the road environment characteristics, determine whether the pedestrian is in a compliant position, and deduct points for violations; extract building distribution information, simulate the line of sight from the driver's perspective, analyze whether the building blocks the pedestrian, and deduct points for violations; Furthermore, the violation points for the traffic sign factor, the violation points for the pedestrian location factor, and the violation points for the surrounding building factor are accumulated and linearly transformed to obtain a road facility association index.

[0013] Collect past traffic accident data from traffic police departments and insurance companies, use web crawler technology to capture photos, videos and descriptions of traffic accident scenes shared by the public from social media and traffic forums, and supplement unofficial records with accident characteristics. The collected traffic accident data can cover various accident cases in different regions, different time periods and multiple traffic scenarios; Furthermore, a multi-layer perceptron model architecture is built, the input layer is designed, the hidden layer is configured, the output layer is set, and the training set, validation set, and test set are divided proportionally; the initial value of the weight is initialized randomly to initialize the weight and bias of the model; the traffic violation risk index, environmental interference intensity index, and road facility association index in the training set are used as input, and the actual proportion of human and natural factors in traffic accidents are used as target outputs, and the input model is used for forward propagation calculations; neurons transmit information in sequence according to weights and activation functions to obtain the predicted output of the model; the difference between the predicted output and the target output is used to calculate the gradient through the back propagation algorithm, and the weights and biases of each layer of the model are updated in reverse; after several rounds of training, the model is finally verified using the test set data, and the model performance is measured using evaluation indicators; and model optimization and adjustment are performed.

[0014] Furthermore, a threshold of the proportion of natural factors is set. When the proportion of natural factors exceeds the threshold, a suggestion for the transformation of the location where the traffic accident occurred is provided based on weather characteristics and road environment characteristics. The method is as follows: Regarding weather characteristics, when it is found in the accident analysis that waterlogging caused by precipitation is the main natural factor, the drainage system of the section where the accident occurred shall be inspected and renovated. In urban roads, drainage outlets shall be set at the prescribed intervals, and the size of the required drainage outlets shall be calculated according to the precipitation and road area to ensure that the drainage capacity meets the requirements; for curved and downhill sections with waterlogging, anti-skid pavement materials shall be used for renovation; when the accident site is affected by strong winds as the main natural factor, it is recommended to install windproof facilities and reinforce road facilities; when the temperature change at the accident site is the main natural factor, it is recommended to upgrade the pavement material and provide vehicle inspection reminders to drivers; when visibility is the main natural factor, it is recommended to improve lighting facilities and install sight guidance facilities; Regarding road environment characteristics, when curve characteristics are the main natural factors, it is recommended to adjust the curve radius, set curve superelevation and take sight improvement measures; when slope characteristics are the main natural factors, it is recommended to adjust the slope and set up auxiliary facilities; when the distribution characteristics of surrounding buildings are the main natural factors, it is recommended to adjust the building setback distance and control building reflections; when the characteristics of road intersections are the main natural factors, it is recommended to optimize traffic control and carry out channelization transformation.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention collects monitoring, meteorological and geographic data in an all-round way, and makes the data interconnected through fine preprocessing and integration. When an accident occurs, the monitoring captures the details of the vehicle collision, while the meteorological data synchronously feedbacks the precise precipitation, wind speed and temperature at that time, and the geographic data provides information on road slope, curves and surrounding buildings.

[0016] 2. The present invention uses YOLOv5 to analyze and monitor, accurately identifying the details of man-made accidents. It not only identifies vehicles that change lanes illegally, exceed speed limits, and fail to follow traffic signs, but also tracks the driver's line of sight and abnormal body movements. It also accurately locates pedestrians' illegal paths and sudden speed changes, making it possible to determine responsibility.

[0017] 3. Through the analysis of past traffic accident data and model training, the present invention can predict in advance the road sections and potential risk factors where accidents may occur; combined with the real-time collected data, the proportion of natural factors is monitored in real time. When it is found that the proportion of natural factors is close to or exceeds the threshold, an early warning is issued in time and corresponding modification suggestions are provided. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a schematic diagram of the steps of the safety warning signal sending control method applied to intelligent monitoring equipment of the present invention; Figure 2 It is a suggested flow chart of the transformation of the traffic accident occurrence site of the safety warning signal sending control method applied to the intelligent monitoring equipment of the present invention. DETAILED DESCRIPTION

[0019] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0020] Example: Figure 1-Figure 2 As shown, the present invention provides a technical solution. According to one embodiment of the present invention, Figure 1 As shown in the schematic diagram of the steps of the safety warning signal sending control method applied to the intelligent monitoring device, the safety warning signal sending control method applied to the intelligent monitoring device includes the following steps: When a traffic accident occurs, obtain monitoring data, meteorological data and geographic data of the accident location, perform data preprocessing and integration, and form traffic accident data; Based on the traffic accident data, the YOLOv5 model is used to extract visual features of targets from the monitoring data, and the targets include vehicles, pedestrians and traffic signs; weather features are extracted based on meteorological data, and precipitation, wind speed, temperature and visibility are quantified as characteristic values ​​of the degree of impact on traffic; road environment features are extracted from geographic data; Based on traffic signs, identify whether the vehicle violates the rules specified by the traffic signs, determine whether the vehicle's driving trajectory complies with the lane rules, and obtain the driver's operation score in combination with weather characteristics to form a traffic violation risk index; combine weather characteristics and road environment characteristics to form an environmental interference intensity index; check the damage of traffic signs around the accident site, obtain the location of pedestrians and the distribution of surrounding buildings, check the obstruction of pedestrians by buildings, and obtain the road facility association index; Collect past traffic accident data, build a multi-layer perceptron model architecture, use the training set data to train the multi-layer perceptron model, continuously adjust the model's weights and biases through the back propagation algorithm, use the traffic violation risk index, environmental interference intensity index, and road facility association index as input, and use the actual proportion of human and natural factors in traffic accidents as the target output; A threshold for the proportion of natural factors is set. When the proportion of natural factors exceeds the threshold, renovation suggestions for the location where the traffic accident occurred are provided based on weather characteristics and road environment characteristics.

[0021] The YOLOv5 model is pre-trained on traffic scene image data. When a traffic accident occurs, the monitoring data in the traffic accident data is input into the YOLOv5 model frame by frame in an image sequence of one frame per second. For vehicles, the model, body color and license plate number are identified, and the driving trajectory of the vehicle before the accident is tracked. For pedestrians, the relative relationship between the pedestrian’s position and the vehicle at the time of the accident is determined. For traffic signs, the sign type and its position in the image are identified.

[0022] Furthermore, based on the weather characteristics, the precipitation is divided into different magnitudes according to meteorological standards, and the depth of water accumulation is estimated in combination with the road environment characteristics. The degree of line of sight obstruction is determined by observing the operating frequency of the wipers, the density of raindrops in front of the camera lens, and the driver's head movement through monitoring data; based on the precipitation magnitude, water accumulation depth, and line of sight obstruction, the precipitation characteristics are quantified; According to the wind speed value in the meteorological data, the corresponding wind force level standard is used. For different wind force level standards and different vehicle models, the impact of wind force on vehicle driving stability is judged and the wind speed characteristics are quantified; Quantify temperature characteristics based on their impact on vehicle components and their association with road conditions; The visibility is classified into different ranges to obtain the visibility range. The driver's operation under different visibility conditions is observed through monitoring data. The influence of the visibility range on the driver's operation is quantified to obtain the visibility characteristics. Furthermore, the quantified precipitation characteristics, wind speed characteristics, temperature characteristics and visibility characteristics are used as weather characteristics.

[0023] Using the geographic information system, the radius of curvature of the curve at the accident site was measured, and combined with the terrain information in the geographic data, it was determined whether there was a slope change at the curve. Based on satellite remote sensing images and field 3D modeling data, the obstruction of the view by vegetation, buildings and terrain undulations around the curve was analyzed. Based on the driver's perspective, the visible range at different positions on the curve was simulated, and the proportion of the obstructed view to the total view was calculated to obtain the curve characteristics. Further, the climbing and downhill sections of the road are identified, the starting and ending points of the climbing or downhill sections are determined according to the altitude change trend in the geographic data, the length and average slope of the climbing and downhill sections are calculated, and the slope characteristics are obtained; Measure the minimum distance between the buildings around the accident site and the road edge, and extract the distribution characteristics of the surrounding buildings; Furthermore, the location of the accident is identified based on the geographic data as to whether it is at an intersection. If so, the type of intersection is further determined, and the traffic lights, signs and markings at the intersection are checked to extract the characteristics of the road intersection. Furthermore, curve characteristics, slope characteristics, surrounding building distribution characteristics and road intersection characteristics are used as road environment characteristics.

[0024] In this embodiment, a two-way four-lane road in the suburbs of the city is selected as the simulated accident section. The section includes a curve with a curvature radius of about 60 meters and a slope of 3%. There are several buildings about 15 meters high around it, about 3 meters away from the edge of the road at the closest point, and a traffic sign with a speed limit of 60km / h is set at the exit of the curve. The experiment simulates a traffic accident under the weather conditions of moderate rain accompanied by level 4 wind and visibility of about 800 meters.

[0025] The YOLOv5 model was used to identify the 60km / h speed limit sign at the exit of the curve. The target tracking algorithm was used to analyze the vehicle's driving trajectory within 100 meters before and after the sign. It was found that a car reached a speed of 75km / h, exceeding the speed limit by 25%, and was recorded as a 5-point violation risk value; at the same time, the car did not turn on the turn signal in advance to change lanes, which was recorded as 2 points. The Hough transform algorithm was used to process the monitoring image sequence to identify clear lane lines. After continuous multi-frame fusion optimization, stable lane line information was obtained. It was observed that another SUV frequently changed lanes 3 times within 10 seconds without turning on the turn signal, and 2 points were recorded each time, for a total of 6 points of violation risk value. In moderate rainy weather, monitoring data showed that the wipers of some vehicles were not turned on normally, affecting the line of sight. The three vehicles involved were recorded as 2 points each; a van slipped on the water-logged road, and the driver did not slow down in time, which was recorded as 3 points. In windy weather, a truck deviated from the normal trajectory due to lateral wind force, and the driver did not adjust the direction, which was recorded as 2 points.

[0026] Adding up the above violation scores, the total violation risk score is: (5+2)+6+(3×2)+3+2=24 points. Using linear transformation, the full score of 30 points corresponds to an index of 10, so the traffic violation risk index = 24 / 30×10=8.

[0027] Moderate rain precipitation causes some water accumulation on the road, with an estimated depth of about 3-4 cm, assigned a value of 3; rain obstructs the line of sight, assigned a value of 3. Level 4 wind has a certain impact on the driving stability of high-body trucks, assigned a value of 2, and there is a risk of falling branches from surrounding trees, assigned a value of 1. The temperature of 20℃ has no obvious impact on vehicle parts, assigned a value of 0. Visibility of 800 meters is slightly restricted, assigned a value of 1, but some drivers did not turn on the fog lights as required, and an additional point was recorded. The comprehensive interference value of weather characteristics is: 3+3+2+1+0+(1+1)=11 points. The radius of curvature of the curve is 60 meters, which is a medium curve, assigned a value of 2, and the slope of 3% is assigned an additional value of 1; the surrounding buildings are close to the road and there is a risk of obstructing the line of sight, assigned a value of 2; there are no special circumstances at the road intersection, assigned a value of 0. The comprehensive interference value of road environment characteristics is: 2+1+2+0=5 points.

[0028] The total environmental interference intensity score is 11+5=16 points after adding up the weather and road environmental interference scores. After linear transformation, the full score of 20 points corresponds to an index of 10, so the environmental interference intensity index = 16 / 20×10=8.

[0029] The speed limit sign was checked and image feature extraction showed that the sign was slightly faded, which resulted in 1 point. Combining GIS data with surveillance footage, it was found that the sign was blocked by roadside branches by about 30% of the field of view, which resulted in 2 points. Traffic sign violation deductions totaled 3 points. Surveillance data tracked a pedestrian crossing the road at a non-pedestrian crossing, which resulted in 2 points. The reflection of surrounding buildings interfered with the driver's vision at a certain angle, which resulted in 1 point. Pedestrian position and building violation deductions totaled 3 points.

[0030] Add up the above violation points, the total violation points = 3 + 3 = 6 points. After linear transformation, the full score of 10 points corresponds to an index of 0, so the road facility correlation index = 6 / 10×10=6.

[0031] Collect past traffic accident data from traffic police departments and insurance companies, use web crawler technology to capture photos, videos and descriptions of traffic accident scenes shared by the public from social media and traffic forums, and supplement unofficial records with accident characteristics. The collected traffic accident data can cover various accident cases in different regions, different time periods and multiple traffic scenarios; Furthermore, a multi-layer perceptron model architecture is built, the input layer is designed, the hidden layer is configured, the output layer is set, and the training set, validation set, and test set are divided proportionally; the initial value of the weight is initialized randomly to initialize the weight and bias of the model; the traffic violation risk index, environmental interference intensity index, and road facility association index in the training set are used as input, and the actual proportion of human and natural factors in traffic accidents are used as target outputs, and the input model is used for forward propagation calculations; neurons transmit information in sequence according to weights and activation functions to obtain the predicted output of the model; the difference between the predicted output and the target output is used to calculate the gradient through the back propagation algorithm, and the weights and biases of each layer of the model are updated in reverse; after several rounds of training, the model is finally verified using the test set data, and the model performance is measured using evaluation indicators; and model optimization and adjustment are performed.

[0032] According to another embodiment of the present invention, Figure 2 As shown in the flowchart of the traffic accident site modification suggestion of the safety warning signal sending control method applied to the intelligent monitoring device, a threshold of the natural factor ratio is set. When the natural factor ratio exceeds the threshold, a modification suggestion of the traffic accident site is provided based on weather characteristics and road environment characteristics. The method is as follows: Regarding weather characteristics, when it is found in the accident analysis that waterlogging caused by precipitation is the main natural factor, the drainage system of the section where the accident occurred shall be inspected and renovated. In urban roads, drainage outlets shall be set at the prescribed intervals, and the size of the required drainage outlets shall be calculated according to the precipitation and road area to ensure that the drainage capacity meets the requirements; for curved and downhill sections with waterlogging, anti-skid pavement materials shall be used for renovation; when the accident site is affected by strong winds as the main natural factor, it is recommended to install windproof facilities and reinforce road facilities; when the temperature change at the accident site is the main natural factor, it is recommended to upgrade the pavement material and provide vehicle inspection reminders to drivers; when visibility is the main natural factor, it is recommended to improve lighting facilities and install sight guidance facilities; Regarding road environment characteristics, when curve characteristics are the main natural factors, it is recommended to adjust the curve radius, set curve superelevation and take sight improvement measures; when slope characteristics are the main natural factors, it is recommended to adjust the slope and set up auxiliary facilities; when the distribution characteristics of surrounding buildings are the main natural factors, it is recommended to adjust the building setback distance and control building reflections; when the characteristics of road intersections are the main natural factors, it is recommended to optimize traffic control and carry out channelization transformation.

[0033] In this embodiment, a section of mountainous and urban junction road in a southern city was selected as the experimental object. The road is about 2 kilometers long, including a 500-meter long continuous curve with a minimum radius of curvature of 60 meters and a slope of 4%. There are some old buildings around the road, about 5 meters away from the road at the closest point. At an intersection on the road, the traffic flow is large.

[0034] Traffic accident data for this road section in the past three years was collected, with a total of 50 accidents. After analysis, it was found that natural factors were more obvious when 20 of the accidents occurred, including weather factors such as precipitation, strong winds, and low visibility, as well as environmental factors such as road bends and slopes. Based on historical data and expert evaluation, the threshold for natural factors was set at 40%.

[0035] After a heavy rainstorm, a three-car rear-end collision occurred on this road section. The precipitation reached 50 mm / hour and lasted for 2 hours. Accurate precipitation data was obtained through the meteorological monitoring station; the process of the accident was recorded using road monitoring video; and geographic data such as the slope and curve of the road were obtained through the Geographic Information System (GIS). After analysis, natural factors accounted for 60% of the accident, mainly due to severe waterlogging on the road caused by precipitation, and the vehicle lost control when passing the curve.

[0036] The drainage system was inspected and renovated, and 15 more drainage outlets were added to the road section, with one outlet every 30 meters. At the same time, the size of the drainage outlet was expanded from the original diameter of 30 cm to 50 cm based on the precipitation and road area. Asphalt pavement materials with better anti-skid performance were used to re-pave the curved and downhill sections with water accumulation.

[0037] In a force 6 gale, a high-body truck was driving on this road section. Due to the lateral wind force, it deviated from the normal driving trajectory and collided with a small car. Meteorological data recorded the wind speed as 12 meters per second; surveillance video recorded the accident; geographic data showed that the section of the road where the accident occurred was relatively empty and there were no effective windbreaks. According to calculations, natural factors accounted for 50% of the accident.

[0038] Two rows of tall poplar trees were planted on both sides of the road as windbreaks, with a tree spacing of 4 meters and a row spacing of 5 meters. Light poles, billboards and other facilities around the road were reinforced, with additional fixed supports and wind-resistant cables.

[0039] A car ran off the road due to excessive speed when passing the curve in this section of road, causing damage to the vehicle and injuries to people. The vehicle's driving trajectory and speed were obtained through monitoring video; the curvature radius and slope of the curve were measured using GIS data. Analysis showed that natural factors accounted for 45% of the accident, mainly because the curve radius was small and the slope was large.

[0040] The curve was modified to expand the radius from 60 meters to 80 meters by excavating part of the mountain and filling it. The curve superelevation was set with a superelevation slope of 3%, making the vehicle more stable when driving on the curve. Some trees that blocked the view were removed around the curve, and a wide-angle mirror was set to improve the driver's view.

[0041] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. A safety warning signal sending control method applied to intelligent monitoring equipment, characterized in that: The method comprises the following steps: When a traffic accident occurs, obtain monitoring data, meteorological data and geographic data of the accident location, perform data preprocessing and integration, and form traffic accident data; Based on the traffic accident data, the YOLOv5 model is used to extract visual features of targets from the monitoring data, and the targets include vehicles, pedestrians and traffic signs; weather features are extracted based on meteorological data, and precipitation, wind speed, temperature and visibility are quantified as characteristic values ​​of the degree of impact on traffic; road environment features are extracted from geographic data; Based on traffic signs, identify whether the vehicle violates the rules specified by the traffic signs, determine whether the vehicle's driving trajectory complies with the lane rules, and obtain the driver's operation score in combination with weather characteristics to form a traffic violation risk index; combine weather characteristics and road environment characteristics to form an environmental interference intensity index; check the damage of traffic signs around the accident site, obtain the location of pedestrians and the distribution of surrounding buildings, check the obstruction of pedestrians by buildings, and obtain the road facility association index; Collect past traffic accident data, build a multi-layer perceptron model architecture, use the training set data to train the multi-layer perceptron model, continuously adjust the model's weights and biases through the back propagation algorithm, use the traffic violation risk index, environmental interference intensity index, and road facility association index as input, and use the actual proportion of human and natural factors in traffic accidents as the target output; A threshold for the proportion of natural factors is set. When the proportion of natural factors exceeds the threshold, renovation suggestions for the location where the traffic accident occurred are provided based on weather characteristics and road environment characteristics.

2. The method for controlling the sending of safety warning signals applied to intelligent monitoring equipment according to claim 1 is characterized in that: Pre-train the YOLOv5 model on traffic scene image data. When a traffic accident occurs, the monitoring data in the traffic accident data is input into the YOLOv5 model frame by frame in an image sequence of one frame per second. For vehicles, the model, body color and license plate number are identified, and the driving trajectory of the vehicle before the accident is tracked. For pedestrians, the relative relationship between the pedestrian’s position and the vehicle at the time of the accident is determined. For traffic signs, identify the sign type and its location in the image.

3. The method for controlling the sending of safety warning signals applied to intelligent monitoring equipment according to claim 2 is characterized in that: Based on the weather characteristics, the precipitation is divided into different levels according to meteorological standards, and the depth of water accumulation is estimated in combination with the road environment characteristics. The degree of line of sight obstruction is determined by observing the operating frequency of the wipers, the density of raindrops in front of the camera lens, and the driver's head movement through monitoring data; based on the precipitation level, water accumulation depth and line of sight obstruction, the precipitation characteristics are quantified; According to the wind speed value in the meteorological data, the corresponding wind force level standard is used. For different wind force level standards and different vehicle models, the impact of wind force on vehicle driving stability is judged and the wind speed characteristics are quantified; Quantify temperature characteristics based on their impact on vehicle components and their association with road conditions; The visibility is classified into different ranges to obtain the visibility range. The driver's operation under different visibility conditions is observed through monitoring data. The influence of the visibility range on the driver's operation is quantified to obtain the visibility characteristics. The quantified precipitation characteristics, wind speed characteristics, temperature characteristics and visibility characteristics are taken as weather characteristics.

4. The method for controlling the sending of safety warning signals applied to intelligent monitoring equipment according to claim 3 is characterized in that: Using the geographic information system, the radius of curvature of the curve at the accident site was measured, and combined with the terrain information in the geographic data, it was determined whether there was a slope change at the curve. Based on satellite remote sensing images and field 3D modeling data, the obstruction of the view by vegetation, buildings and terrain undulations around the curve was analyzed. Based on the driver's perspective, the visible range at different positions on the curve was simulated, and the proportion of the obstructed view to the total view was calculated to obtain the curve characteristics. Identify the climbing and downhill sections of the road, determine the starting and ending points of the climbing or downhill sections according to the altitude change trend in the geographic data, calculate the length and average slope of the climbing and downhill sections, and obtain the slope characteristics; Measure the minimum distance between the buildings around the accident site and the road edge, and extract the distribution characteristics of the surrounding buildings; Identify whether the accident site is at an intersection based on geographic data. If so, further determine the type of intersection, check the traffic lights, signs and markings at the intersection, and extract road intersection features; Curve characteristics, slope characteristics, surrounding building distribution characteristics and road intersection characteristics are taken as road environment characteristics.

5. The method for controlling the sending of safety warning signals applied to intelligent monitoring equipment according to claim 1 is characterized in that: Identify and locate traffic signs, and based on traffic rules, determine vehicle violations and deduct points for violations; use the Hough transform algorithm to identify the position and shape of lane lines, and obtain lane line information by fusing and optimizing the lane line detection results of multiple consecutive frames of images; track the vehicle's driving trajectory on the road based on the lane line information, use the target tracking algorithm to record the vehicle's center position, driving direction, and relative position relationship with the lane line, and determine the vehicle's violations based on traffic rules and deduct points for violations; When the meteorological data shows that the precipitation intensity reaches moderate rain or above, the driver's use of wipers will be observed through monitoring data to determine whether the rain has obstructed the driver's vision, and points will be deducted for violations; the driving stability of the vehicle will be observed. If the vehicle slips due to water on the road and the driver does not take measures to slow down or avoid it, points will be deducted for violations; operational considerations in weather with insufficient visibility and driving behavior analysis in windy weather will be conducted, and points will be deducted for violations; The violation points are accumulated to obtain the total violation risk score, and a linear change is used to obtain the traffic violation risk index.

6. The method for controlling the sending of safety warning signals applied to intelligent monitoring equipment according to claim 1 is characterized in that: According to the precipitation and precipitation intensity information in the meteorological data, the degree of interference to traffic is quantified; according to the wind speed value in the meteorological data, the interference intensity is judged according to the wind force level standard; according to the temperature in the meteorological data, the degree of impact on automobile parts is judged; the visibility is graded and quantified according to different ranges; Quantify interference based on road environment characteristics and evaluate the interference values ​​of curve characteristics, slope characteristics, surrounding building distribution characteristics and road intersection characteristics; The various interference values ​​based on weather characteristics and the various interference values ​​based on road environment characteristics are accumulated to obtain the total environmental interference intensity score; and the environmental interference intensity index is obtained by adopting linear change.

7. The method for controlling the sending of safety warning signals applied to intelligent monitoring equipment according to claim 1 is characterized in that: For each identified traffic sign, image feature extraction is used to determine whether its surface is damaged, faded, or deformed, and the damage score of the traffic sign is obtained. The source of obstructions around the traffic sign is determined by combining geographic information system data and monitoring images, and the obstruction score of the traffic sign is quantified by measuring the proportion of the obstructions in the field of view directly in front of the traffic sign. The damage score and the obstruction score of the traffic sign are added together to determine the violation points for the traffic sign factor.

8. The method for controlling the sending of safety warning signals applied to intelligent monitoring equipment according to claim 7 is characterized in that: Based on monitoring data, the target tracking algorithm is used to lock pedestrians at the accident scene and surrounding areas in real time, match the pedestrian's location with the road environment characteristics, determine whether the pedestrian is in a compliant position, and deduct points for violations; Extract building distribution information, simulate sight lines from the driver's perspective, analyze whether buildings block pedestrians, and deduct points for violations; The violation points for the traffic sign factor, the violation points for the pedestrian location factor, and the violation points for the surrounding building factor are accumulated and linearly transformed to obtain a road facility association index.

9. The method for controlling the sending of safety warning signals applied to intelligent monitoring equipment according to claim 1 is characterized in that: Collect past traffic accident data from traffic police departments and insurance companies, use web crawler technology to capture photos, videos and descriptions of traffic accident scenes shared by the public from social media and traffic forums, and supplement unofficial records with accident characteristics. The collected traffic accident data can cover various accident cases in different regions, different time periods and multiple traffic scenarios; Build a multi-layer perceptron model architecture, design the input layer, configure the hidden layer, set the output layer, and divide the training set, validation set, and test set in proportion; use the random initialization method for the initial value of the weight to initialize the weight and bias of the model; use the traffic violation risk index, environmental interference intensity index, and road facility association index in the training set as input, and the actual proportion of human and natural factors in traffic accidents as the target output, and input the model for forward propagation calculation; neurons transmit information in sequence according to weights and activation functions to obtain the predicted output of the model; use the difference between the predicted output and the target output to calculate the gradient through the back propagation algorithm, and reversely update the weights and biases of each layer of the model; after several rounds of training, use the test set data to finally verify the model, and use evaluation indicators to measure the model performance; perform model optimization and adjustment.

10. The method for controlling the sending of safety warning signals applied to intelligent monitoring equipment according to claim 9, characterized in that: Set a threshold for the proportion of natural factors. When the proportion of natural factors exceeds the threshold, provide renovation suggestions for the location where the traffic accident occurred based on weather characteristics and road environment characteristics. The method is as follows: Regarding weather characteristics, when it is found in the accident analysis that waterlogging caused by precipitation is the main natural factor, the drainage system of the section where the accident occurred shall be inspected and renovated. In urban roads, drainage outlets shall be set at the prescribed intervals, and the size of the required drainage outlets shall be calculated according to the precipitation and road area to ensure that the drainage capacity meets the requirements; for curved and downhill sections with waterlogging, anti-skid pavement materials shall be used for renovation; when the accident site is affected by strong winds as the main natural factor, it is recommended to install windproof facilities and reinforce road facilities; when the temperature change at the accident site is the main natural factor, it is recommended to upgrade the pavement material and provide vehicle inspection reminders to drivers; when visibility is the main natural factor, it is recommended to improve lighting facilities and install sight guidance facilities; Regarding road environment characteristics, when curve characteristics are the main natural factors, it is recommended to adjust the curve radius, set curve superelevation and take sight improvement measures; when slope characteristics are the main natural factors, it is recommended to adjust the slope and set up auxiliary facilities; when the distribution characteristics of surrounding buildings are the main natural factors, it is recommended to adjust the building setback distance and control building reflections; when the characteristics of road intersections are the main natural factors, it is recommended to optimize traffic control and carry out channelization transformation.

Images