Heavy-duty truck brake out-of-control state identification and warning system and method

Abstract

The invention belongs to the technical field of road traffic active safety, and particularly relates to a heavy-duty truck braking out-of-control state identification and warning system and method.A hub temperature monitoring sub-module in a data monitoring module is used for obtaining truck information and hub temperature and judging whether the truck information and the hub temperature exceed an alarm value or not; a driving state monitoring sub-module in the data monitoring module is used for obtaining the truck information and vehicle operation characteristics and sending the truck information and the vehicle operation characteristics to the edge calculation module; the edge calculation module is used for receiving and integrating the data sent by the data monitoring module, calling an out-of-control state identification model based on a deep learning algorithm to judge the state of the vehicle, and generating warning information according to a judgment result; and the out-of-control accident warning module is used for receiving the warning information and warning out an out-of-control accident. The problems that the out-of-control state of the heavy-duty truck is difficult to monitor, misinformation exists in monitoring, and warning is delayed are solved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of highway traffic active safety, in particular to a heavy truck brake out-of-control state identification and warning system and method. BACKGROUND

[0002] With the continuous growth of highway freight volume, heavy trucks play an important role in transportation. However, due to the characteristics of large vehicle load, large inertia, high load of braking system, etc., heavy trucks are prone to brake out-of-control due to brake thermal recession or mechanical failure under long downhill working conditions, and thus cause serious traffic accidents.

[0003] Currently, for the identification and warning of truck brake out-of-control state, the existing technology mainly realizes it in the following ways: (1) threshold judgment based on brake system pressure, triggering an alarm when the detected value exceeds the preset threshold; (2) using a vehicle-mounted temperature sensor to detect the hub temperature, triggering an alarm when the hub temperature exceeds the set threshold; (3) inferring the out-of-control state through vehicle acceleration or speed change rate; (4) combining GPS positioning and map data to provide risk warning for trucks driving on long downhill sections. However, the above methods have obvious defects: first, the judgment relying on a single parameter is easily disturbed, has a high false alarm rate, and is difficult to reflect the gradual failure of the braking system in time; second, traditional acceleration detection has poor adaptability under complex road conditions (such as bumps and curves); in addition, the existing warning systems are mostly limited to in-vehicle alarms or use roadside warning signs to remind of the out-of-control risk, lack of collaborative warning capability for surrounding vehicles and road management systems, and are difficult to provide sufficient reaction time for risk avoidance decisions. SUMMARY

[0004] The technical problem solved by the present application is to provide a heavy truck brake out-of-control state identification and warning system and method to actively monitor the out-of-control state and solve the problems of false alarm, missed alarm and delayed warning in heavy truck out-of-control state monitoring.

[0005] The present application provides a basic scheme: a heavy truck brake out-of-control state identification and warning system, characterized in that it comprises The data monitoring module comprises a wheel hub temperature monitoring submodule and a driving state monitoring submodule; the wheel hub temperature monitoring submodule is configured to acquire truck information and wheel hub temperature, determine whether the truck wheel hub temperature exceeds an alarm value, and if so, determine that the current truck is a brake out-of-control risk truck, and send the truck information and wheel hub temperature of the truck to the edge computing module; the driving state monitoring submodule is configured to acquire truck information, vehicle operating characteristics, and vehicle driving trajectory, and send them to the edge computing module, wherein the vehicle operating characteristics comprise speed, acceleration, and heading angle, and the truck information comprises vehicle type and license plate; the edge computing module is configured to receive and integrate the monitoring information sent by the data monitoring module, and call a deep learning algorithm-based out-of-control state identification model to determine whether the vehicle is in an out-of-control state, and is further configured to generate warning information according to the vehicle out-of-control state determination result and vehicle driving trajectory, wherein the vehicle type, license plate, acceleration, speed, heading angle, and wheel hub temperature are used as model input parameters, and the vehicle out-of-control probability is used as model output. The out-of-control accident warning module comprises a display unit configured to receive the warning information and warn the out-of-control accident according to the warning information in a preset warning mode.

[0006] Preferably, the wheel hub temperature monitoring submodule comprises a non-contact infrared sensor arranged on the roadside, and the driving state monitoring submodule comprises a radar multi-dimensional intelligent monitoring all-in-one machine arranged on the roadside.

[0007] Preferably, the out-of-control state identification model comprises an input layer, a self-attention mechanism layer, a deep residual network layer, and an output layer, the input layer is configured to integrate the vehicle type, license plate, acceleration, speed, heading angle, and brake drum temperature data; the self-attention mechanism layer is configured to calculate the attention weight of each feature data and highlight the features related to the out-of-control state; the deep residual network layer is configured to extract the deep relationship between the features and calculate the vehicle out-of-control probability; and the output layer is configured to output the vehicle out-of-control probability, wherein the vehicle out-of-control probability comprises the vehicle out-of-control state probability and the vehicle normal state probability.

[0008] Further preferably, the self-attention mechanism layer calculates the attention weight by the following formula:

[0009] wherein Q, K, and V are query vector, key vector, and value vector, respectively, is the vector dimension.

[0010] Further preferably, the deep residual network layer comprises a convolution layer, a pooling layer, and a fully connected layer, and the deep residual network layer operates as follows:

[0011]

[0012]

[0013] wherein, is the result of convolution, is the weight matrix, is the input data block, b is the bias vector in the convolution process, f is the activation function, c is the complete feature map, n is the length of the input data, h is the size of the convolution kernel, n-h+1 is the length of the output feature map, is the reduced dimension feature map after pooling, m is the length of the feature map after pooling, is the output value of the jth window of the pooling operation.

[0014] Preferably, the pre-warning value is 250-310 DEG C.

[0015] Preferably, the warning mode is at least one of a gantry information board, an F-shaped information board and a T-shaped information board, which sends warning information and risk avoidance guidance information of the out-of-control vehicle to other vehicles.

[0016] Preferably, the edge computing module comprises a data processing submodule, an identification model submodule and a warning information generation submodule, The data processing submodule is used to integrate the information sent by the wheel hub temperature monitoring submodule and the driving state monitoring submodule according to the truck information, and form a data information list of the brake out-of-control risk truck. The identification model submodule is used to input the vehicle model, license plate, acceleration, speed, heading angle and wheel hub temperature of the brake out-of-control risk truck, and output the vehicle out-of-control probability through the out-of-control state identification model. The warning information generation submodule is used to compare the out-of-control state probability output by the out-of-control state identification model with the out-of-control probability threshold value, and determine the out-of-control truck if the threshold value is exceeded; and is also used to generate the warning information corresponding to each out-of-control accident warning module according to the truck information, vehicle operation characteristics, vehicle driving track and distribution position of the out-of-control accident warning module of the out-of-control truck, and send the warning information to the corresponding out-of-control accident warning module.

[0017] Another basic scheme provided by the application is a method for identifying and warning the brake out-of-control state of a heavy truck, comprising the following steps: S1: obtaining truck information and truck wheel hub temperature, judging whether the truck wheel hub temperature exceeds a pre-warning value, determining the current truck as a brake out-of-control risk truck if the pre-warning value is exceeded, obtaining the vehicle operation characteristics of the truck, and sending the truck information, truck wheel hub temperature and vehicle operation characteristics of the brake out-of-control risk truck to the edge computing, wherein the vehicle operation characteristics include speed, acceleration and heading angle, and the truck information includes license plate and vehicle model. S2: collate the truck information and the truck hub temperature, vehicle running feature information, input the vehicle model, license plate, acceleration, speed, heading angle and hub temperature into the vehicle out-of-control state identification model, and output the vehicle out-of-control probability; S3: generating warning information according to the vehicle out-of-control probability, and warning the out-of-control accident according to a preset warning mode.

[0018] The principle and advantages of the present application are: 1. The present application combines hub temperature and vehicle driving features, considers out-of-control state judgment indicators comprehensively, reduces judgment errors caused by single data source, and dynamically weights multiple features through the out-of-control state identification model based on deep learning to accurately identify the out-of-control state, multi-modal data fusion judgment, and improve the judgment accuracy. Compared with the judgment method relying on only a single threshold, the accuracy of the judgment is greatly improved.

[0019] 2. The edge device deploys a lightweight model, processes data locally, avoids network delay, meets the real-time demand of out-of-control early warning (such as responding within 10ms on a long downhill section), and guarantees real-time performance.

[0020] 3. Hierarchical judgment, first judging whether the monitored current truck is a brake out-of-control risk truck according to the hub temperature, then sending the hub temperature, truck information and vehicle driving features corresponding to the truck to the edge computing for further judgment by the out-of-control state identification model, and issuing a warning according to the judgment result. Hierarchical judgment reduces the calculation amount of edge computing, and the out-of-control state identification model can accurately judge the truck that may exist out-of-control risk, so as to take measures to avoid the out-of-control risk. BRIEF DESCRIPTION OF DRAWINGS

[0021] Figure 1 is a logic block diagram of the system of the present application; Figure 2 is a flowchart of the method of the present application. DETAILED DESCRIPTION

[0022] The following will be further described in detail through specific embodiments: The specific implementation process is as follows: Embodiment 1 Referring to Figure 1 A heavy truck brake out-of-control state identification and warning system, which comprises The data monitoring module comprises a wheel hub temperature monitoring submodule and a driving state monitoring submodule; the wheel hub temperature monitoring submodule is configured to acquire truck information and wheel hub temperature, determine whether the wheel hub temperature of the truck exceeds an alarm value, and if so, determine that the current truck is a brake out-of-control risk truck, and send the truck information and the wheel hub temperature of the truck to the edge computing module; the driving state monitoring submodule is configured to acquire truck information, vehicle operating characteristics and vehicle driving trajectory, and send them to the edge computing module, wherein the vehicle operating characteristics comprise speed, acceleration and heading angle, and the truck information comprises vehicle type and license plate; The wheel hub temperature monitoring submodule comprises a non-contact infrared sensor arranged on the roadside of the road, specifically, the non-contact infrared sensor is arranged at the same height as the wheel hub of the truck, the angle range of the wide-angle infrared sensor can reach 60° to 120°, the wheel hub of the truck can be monitored, and multiple target objects can be monitored at one time. The non-contact infrared sensor is configured to collect a truck brake wheel hub infrared image, and the wheel hub temperature monitoring submodule analyzes the highest temperature of the wheel hub according to the truck brake wheel hub infrared image, and takes the temperature as the wheel hub temperature. The driving state monitoring submodule comprises a truck information recognition unit, a vehicle operating characteristic acquisition unit and a vehicle operating trajectory generation unit. The truck information recognition unit acquires a vehicle image through a camera, and identifies the license plate and vehicle type of the vehicle through analyzing the vehicle image, and if the vehicle is a truck, the identified license plate and vehicle type are taken as the truck information. The vehicle operating characteristic acquisition unit acquires the vehicle operating characteristics, i.e. acceleration, speed and heading angle, of the vehicle identified as a truck through a high-frequency millimeter wave radar. The vehicle operating trajectory generation unit generates a vehicle operating trajectory according to the vehicle image and the vehicle operating characteristics of the truck.

[0023] The pre-warning value is 250-310℃, and the pre-warning value in the example is 300℃.

[0024] Specifically, in the example, the wheel hub temperature monitoring submodule and the driving state monitoring submodule are arranged at intervals of 600 meters on a long downhill section, and the non-contact infrared sensor adopts a ThermoView TV40 series sensor. The camera of the driving state monitoring submodule adopts an intelligent high-definition camera, and in the example, a Hikvision iDS-TCS802-HS-L camera is adopted.

[0025] The edge computing module is configured to receive and integrate the monitoring information sent by the data monitoring module, call a deep learning algorithm-based out-of-control state identification model to determine whether the vehicle is in an out-of-control state, and generate warning information according to the vehicle out-of-control state determination result and the vehicle driving trajectory, wherein the vehicle type, license plate, acceleration, speed, heading angle and wheel hub temperature are taken as model input parameters, and the vehicle out-of-control probability is taken as model output. The edge computing module comprises a data processing submodule, an identification model submodule, and an alarm information generation submodule. The data processing submodule is configured to integrate information sent by the wheel hub temperature monitoring submodule and the driving state monitoring submodule according to the truck information, and form a data information list of the brake out-of-control risk truck. The identification model submodule is configured to input the vehicle model, the license plate, the acceleration, the speed, the heading angle, and the wheel hub temperature of the brake out-of-control risk truck, and output a vehicle out-of-control probability through the out-of-control state identification model. The alarm information generation submodule is configured to compare the vehicle out-of-control state probability output by the out-of-control state identification model with a threshold value of the out-of-control probability, and determine that the truck is out of control if the threshold value is exceeded. The alarm information generation submodule is also configured to generate alarm information corresponding to each out-of-control accident alarm module according to the truck information, the vehicle operation characteristics, the vehicle driving track, and the distribution position of the out-of-control accident alarm module of the out-of-control truck, and send the alarm information to the corresponding out-of-control accident alarm module. The alarm information comprises the current position of the out-of-control vehicle, the license plate number, the speed, the distance from the out-of-control accident alarm module, the position of the safe lane, and the like, and is used to guide other vehicles to avoid and guide the out-of-control vehicle. The threshold value of the out-of-control probability is 0.65-0.75, and the threshold value of the out-of-control probability in the embodiment is 0.7. The distribution position of the out-of-control accident alarm module is stored in the memory of the edge computing module and is updated in real time according to the actual situation.

[0026] The out-of-control accident alarm module is in communication connection with the edge computing module. The module comprises a display unit configured to receive the alarm information and alarm the out-of-control accident according to the alarm information in a preset alarm mode. The communication mode between the out-of-control accident alarm module and the edge computing module is 5G.

[0027] The out-of-control state identification model comprises an input layer, a self-attention mechanism layer, a deep residual network layer, and an output layer. The input layer is configured to integrate the vehicle model, the license plate, the acceleration, the speed, the heading angle, and the brake drum temperature data. The self-attention mechanism layer is configured to calculate the attention weight of each feature data and highlight the features related to the out-of-control state. The deep residual network layer is configured to extract the deep relationship between the features and calculate the vehicle out-of-control probability. The output layer is configured to output the vehicle out-of-control probability, which comprises the vehicle out-of-control state probability and the vehicle normal state probability. The number of neurons of the input layer is 6, and the number of neurons of the output layer is 2.

[0028] The self-attention mechanism layer calculates the attention weight through the following formula:

[0029] Q, K, and V are respectively a query vector, a key vector, and a value vector, is the vector dimension.

[0030] The deep residual network layer includes a convolution layer, a pooling layer and a full connection layer, and the deep residual network layer operates as follows:

[0031]

[0032]

[0033] wherein, is the result of convolution, is a weight matrix, is an input data block, b is a bias vector in the convolution process, f is an activation function, c is a complete feature map, n is the length of the input data, h is the size of the convolution kernel, n-h+1 is the length of the output feature map, is a reduced dimension feature map after pooling, m is the length of the feature map after pooling, is the output value of the jth window of the pooling operation.

[0034] The output layer includes two neurons, respectively representing the probability of the truck being in an out-of-control state and a normal driving state. The output result can be represented by a vector , for example, out-of-control normal, wherein represents the probability of the truck being out of control, represents the probability of the truck being in a normal driving state.

[0035] The warning mode is at least one of a portal information board, an F-shaped information board and a T-shaped information board, which sends warning information and risk avoidance guidance information of the out-of-control vehicle to other vehicles.

[0036] Embodiment 2 Referring to Figure 2 , a heavy truck brake out-of-control state identification and warning method is used for the above system, including the following steps: S1: Obtain truck information and truck hub temperature, and determine whether the truck hub temperature exceeds the warning value. If it exceeds the warning value, it is determined that the current truck is a brake out-of-control risk truck, and the vehicle running characteristics and vehicle driving track of the truck are obtained. The truck information and truck hub temperature, vehicle running characteristics and vehicle driving track of the brake out-of-control risk truck are sent to the edge calculation. The vehicle running characteristics include speed, acceleration and heading angle. The truck information includes license plate and vehicle type.

[0037] S2: collate the truck information and the truck hub temperature, the vehicle running characteristics, the vehicle driving track information, input the vehicle model, the license plate, the acceleration, the speed, the heading angle and the hub temperature into the truck out-of-control state identification model, and output the vehicle out-of-control probability; the truck out-of-control state identification model is a model based on a deep learning algorithm, and specifically adopts the identification model of embodiment 1.

[0038] S3: generate warning information according to the vehicle out-of-control probability and the vehicle driving track, and warn the out-of-control accident according to a preset warning mode.

[0039] The above is only an embodiment of the present application, and common knowledge such as specific structures and characteristics in the scheme is not described in detail here. The person skilled in the art knows all the common technical knowledge in the field of the present application before the filing date or the priority date, can know all the prior art in the field, and has the ability to apply conventional experimental means before that date. The person skilled in the art can improve and implement the present scheme based on the disclosure given in the present application, and their own ability. Some typical known structures or known methods should not be an obstacle for the person skilled in the art to implement the present application. It should be noted that for those skilled in the art, without departing from the structure of the present application, a number of modifications and improvements can be made, which should also be considered within the scope of protection of the present application. These will not affect the effect and practicality of the present application. The scope of protection claimed in the present application should be subject to the content of its claims, and the specific implementation mode and the like in the specification can be used to explain the content of the claims.

Claims

1. A heavy-duty truck brake out-of-control state identification and warning system, characterized in that: include The data monitoring module includes a wheel hub temperature monitoring submodule and a driving status monitoring submodule. The wheel hub temperature monitoring submodule is used to obtain truck information and wheel hub temperature, determine whether the truck wheel hub temperature exceeds the alarm value, and if so, determine that the current truck is a truck with a brake loss risk, and send the truck information and wheel hub temperature to the edge computing module. The driving status monitoring submodule is used to obtain truck information, vehicle operation characteristics, and vehicle driving trajectory, and send them to the edge computing module. The vehicle operation characteristics include speed, acceleration, and heading angle. The truck information includes vehicle model and license plate. The edge computing module is used to receive and integrate monitoring information sent by the data monitoring module, and call the out-of-control state identification model based on the deep learning algorithm to determine whether the vehicle is in an out-of-control state. It is also used to generate warning information based on the vehicle out-of-control state judgment result and the vehicle's driving trajectory. The model uses the vehicle model, license plate, acceleration, speed, heading angle, and wheel hub temperature as model input parameters, and the vehicle out-of-control probability as the model output; The out-of-control accident warning module includes a display unit, which is used to receive warning information and warn of the out-of-control accident in a preset warning manner according to the warning information.

2. The heavy-duty truck brake out-of-control state identification and warning system according to claim 1 is characterized in that: The wheel hub temperature monitoring submodule includes a non-contact infrared sensor arranged on the side of the road, and the driving status monitoring submodule includes a Leishi multi-dimensional intelligent monitoring integrated machine arranged on the side of the road.

3. The heavy truck brake out-of-control state identification and warning system according to claim 1 is characterized in that: The out-of-control state identification model includes an input layer, a self-attention mechanism layer, a deep residual network layer, and an output layer. The input layer is used to integrate vehicle model, license plate, acceleration, speed, heading angle and brake drum temperature data; the self-attention mechanism layer is used to calculate the attention weight of each feature data and highlight the features related to the out-of-control state; the deep residual network layer is used to extract the deep-level relationship between features and calculate the vehicle out-of-control probability; the output layer is used to output the vehicle out-of-control probability, which includes the vehicle out-of-control state probability and the vehicle normal state probability.

4. The heavy-duty truck brake out-of-control state identification and warning system according to claim 3 is characterized by: The self-attention mechanism layer calculates the attention weight using the following formula: Among them, Q, K, and V are query vector, key vector, and value vector respectively. is the vector dimension.

5. The heavy truck brake out-of-control state identification and warning system according to claim 3 is characterized in that: The deep residual network layer includes a convolutional layer, a pooling layer, and a fully connected layer. The operation of the deep residual network layer is as follows: in, is the result of convolution, is the weight matrix, is the input data block, b is the bias vector in the convolution process, f is the activation function, c is the complete feature map, n is the length of the input data, h is the size of the convolution kernel, and n-h+1 is the length of the output feature map. is the dimension reduction feature map after pooling, m is the length of the feature map after pooling, is the output value of the j-th window of the pooling operation.

6. The heavy-duty truck brake out-of-control state identification and warning system according to claim 1 is characterized in that: The warning value is 250-310°C.

7. The heavy-duty truck brake out-of-control state identification and warning system according to claim 1, characterized in that: The warning method is through at least one of a gantry-type information board, an F-type information board and a T-type information board, and each information board displays out-of-control vehicle avoidance guidance information and other driving avoidance information.

8. The heavy truck brake out-of-control state identification and warning system according to claim 1 is characterized in that: The edge computing module includes a data processing submodule, an identification model submodule, and a warning information generation submodule. The data processing submodule is used to integrate the information sent by the wheel hub temperature monitoring submodule and the driving status monitoring submodule according to the truck information to form a data information list of trucks with brake loss risk; The identification model submodule is used to input the vehicle model, license plate, acceleration, speed, heading angle and wheel hub temperature of the truck with brake out-of-control risk, and output the vehicle out-of-control probability through the out-of-control state identification model; The warning information generation submodule is used to compare the vehicle out-of-control state probability output by the out-of-control state identification model with the out-of-control probability threshold, and if the probability exceeds the threshold, it is determined to be an out-of-control truck; it is also used to generate warning information corresponding to each out-of-control accident warning module based on the truck information, vehicle operation characteristics, vehicle driving trajectory, and distribution location of the out-of-control accident warning module, and send the warning information to the corresponding out-of-control accident warning module.

9. A method for identifying and warning a brake failure state of a heavy-duty truck according to any one of claims 1 to 8, characterized in that: The following steps are involved: S1: Obtain truck information and wheel hub temperature, determine whether the wheel hub temperature exceeds the warning value, and if so, determine that the current truck is a truck with a brake loss risk. Obtain the vehicle operation characteristics of the truck and send the truck information, wheel hub temperature, and vehicle operation characteristics of the truck with a brake loss risk to edge computing. The vehicle operation characteristics include speed, acceleration, and heading angle, and the truck information includes license plate and vehicle model. S2: Organize truck information, wheel hub temperature, and vehicle operating characteristics, input vehicle model, license plate, acceleration, speed, heading angle, and wheel hub temperature into the truck out-of-control state identification model, and output the vehicle out-of-control probability; S3: Generate warning information based on the probability of vehicle loss of control, and issue a warning for the loss of control accident according to the preset warning method.

Images