Vehicle monitoring system based on ECU data integration and intelligent management

Abstract

The invention discloses a vehicle monitoring system based on ECU (Electronic Control Unit) data integration and intelligent management, which comprises an ECU data reading module, an intelligent data integration module, a Bluetooth communication module, an accident real-time detection module, a remote supervision module and a driver feedback module, the intelligent data integration module is used for intelligent integration and management of ECU data, the Bluetooth communication module is used for achieving independent voice reminding of a driver, the accident real-time detection module is used for automatically recognizing traffic accidents and storing on-site videos, and the remote supervision module is used for transmitting vehicle data to a supervision platform. According to the vehicle monitoring system based on ECU data integration and intelligent management, a data optimization integration algorithm based on multi-modal data fusion is provided to carry out intelligent integration on multi-modal vehicle data; and proposing a real-time traffic accident identification algorithm based on abnormal motion detection to automatically identify the traffic accident.

Description

Technical Field

[0001] The present invention relates to the fields of multimodal data fusion, intelligent management and anomaly detection, and in particular to a vehicle monitoring system based on ECU data integration and intelligent management. Background Art

[0002] Multimodal data fusion technology is an intelligent technology used to integrate multimodal data from different sensors and vehicle ECUs. It aims to solve the data inconsistency and redundancy problems caused by the diversity of sensor data sources. By optimizing and integrating multimodal data sources under the vehicle's operating state, the system's high fault tolerance and data reliability can be achieved. Multimodal data fusion technology can improve data processing efficiency and provide accurate basic data support for the system's intelligent management and decision-making.

[0003] Intelligent management technology is a data-driven management technology that aims to solve the problems of inefficient management and inaccurate decision-making in vehicle monitoring systems. Through intelligent analysis of vehicle operating status, driving behavior and sensor data, intelligent management technology can optimize vehicle management processes and provide real-time alarm and data analysis support.

[0004] Abnormal motion detection technology is an automatic accident identification technology based on intelligent algorithms. It aims to solve the problems of delayed accident response and inaccurate detection in traditional traffic monitoring systems. Abnormal motion detection technology continuously monitors the vehicle's motion data and can identify abnormal movement patterns, including sudden braking and lateral movement. When a potential accident is detected, it automatically triggers the system to record and save real-time video.

[0005] However, the existing vehicle monitoring system based on ECU data integration and intelligent management has weak data integration capabilities from different sensors and is difficult to effectively process large amounts of heterogeneous data, resulting in incomplete coverage of vehicle operating status information. Secondly, the existing monitoring system relies on manual operation and has a low degree of automation. It cannot achieve real-time automatic identification and alarm in traffic accident detection. In addition, the system has a slow response speed when dealing with emergencies and is unable to quickly obtain and save key data at the accident scene, affecting subsequent accident responsibility determination and data analysis. Summary of the invention

[0006] The purpose of the present invention is to provide a vehicle monitoring system based on ECU data integration and intelligent management, so as to solve the problems that the existing vehicle monitoring system based on ECU data integration and intelligent management proposed in the above background technology has weak data integration capability from different sensors, and is difficult to effectively process a large amount of heterogeneous data, resulting in incomplete coverage of vehicle operation status information. Secondly, the existing monitoring system relies on manual operation and has a low degree of automation. In terms of traffic accident detection, it is impossible to achieve real-time automatic identification and alarm. In addition, the system has a slow response speed when dealing with sudden accidents, and is unable to quickly obtain and save key data at the accident scene, affecting subsequent accident responsibility determination and data analysis.

[0007] To achieve the above-mentioned purpose, the present invention provides the following technical solutions: a vehicle monitoring system based on ECU data integration and intelligent management, comprising an ECU data reading module, an intelligent data integration module, a Bluetooth communication module, a real-time accident detection module, a remote supervision module, and a driver feedback module, characterized in that: the ECU data reading module is used to obtain the vehicle's operating status data in real time, including engine data, transmission system data, vehicle dynamic data, body control data, and environmental perception data, to ensure that the system can continuously monitor and provide accurate vehicle operation information; the intelligent data integration module comprises an ECU data integration unit and a data intelligent management unit; the ECU data integration unit proposes a data optimization integration algorithm based on multimodal data fusion for intelligently integrating multimodal data information read from the ECU, and provides comprehensive vehicle operating status information by optimizing and integrating data from different sensors; the data The intelligent management unit is used to intelligently manage the integrated ECU data, analyze the vehicle's maintenance needs and driving behavior information, and ensure the long-term and efficient operation of the vehicle. The Bluetooth communication module is used to implement individual voice reminders for the driver, and provide vehicle status and accident warning information in a timely manner to ensure that the driver focuses on driving and obtains necessary real-time information. The real-time accident detection module proposes a real-time traffic accident recognition algorithm based on abnormal motion detection to automatically identify the occurrence of traffic accidents. When an accident is detected, the camera is automatically triggered and the on-site video is saved to facilitate subsequent analysis and responsibility definition. The remote supervision module is used to transmit the vehicle's operating data to the supervision platform in real time through the Beidou system to ensure the safety and compliance of the vehicle. The driver feedback module is used for interactive feedback between the system and the driver. By providing operational suggestions, real-time data reports and post-accident processing solutions, it helps the driver to efficiently understand the vehicle's status and improve driving safety.

[0008] Preferably, the ECU data reading module obtains the vehicle's operating status data in real time through the vehicle's hardware interface, including speed, engine status, and fuel consumption data, to ensure that the system can fully grasp the vehicle's operating status, provide real-time vehicle information, and support subsequent data processing and monitoring functions.

[0009] Preferably, the intelligent data integration module includes an ECU data integration unit, which proposes a data optimization integration algorithm based on multimodal data fusion for intelligently integrating multimodal data information read from the ECU, ensuring that different types of data can be processed uniformly and providing comprehensive vehicle operating status information.

[0010] Specifically, the data optimization and integration algorithm based on multimodal data fusion is as follows: First, an ECU data fusion model is established. The original ECU data from different sensors can be integrated through a centralized ECU data architecture, including engine ECU data, transmission system ECU data, vehicle dynamics ECU data, body control ECU data, and environmental perception ECU data, to ensure that all sensor ECU data can be processed on a unified platform for subsequent ECU data optimization and integration. The specific formula is expressed as follows:

[0011] Y=HX+V

[0012] Among them, Y represents the fused observed ECU data, H represents the observation matrix, which represents the mapping of different sensor ECU data to the observation space, X represents the system state vector, which contains the state information of all sensors, and V represents the noise matrix. Secondly, the Kalman filter is used for state estimation and prediction to optimize and integrate ECU data from different sensors. The specific formula of the constructed dynamic prediction model is expressed as follows:

[0013]

[0014] in, It is represented as the predicted estimate of the system state at time k based on the information at time k-1. It represents the prior state estimate, which means the best guess of the system state before receiving the observed ECU data at time k. A is represented as the state transfer matrix, which represents the matrix of the evolution of the system state over time. The estimation of the system state at time k-1 based on the information at time k-1 represents the a posteriori state estimation, which represents the best guess of the system state after receiving the observed ECU data at time k. B is represented as the control input matrix. The control input U k Associated with state transfer, U kIt is represented as the control input information applied at time k, W is represented as the process noise matrix, which represents the random variables in the state transfer process. W is Gaussian distributed and has zero mean and covariance matrix Q. k is represented as the discrete time index. By building a dynamic model to reduce the impact of sensor noise and improve the accuracy of ECU data, the constructed dynamic prediction model is then updated to use the new observed ECU data to correct the previously predicted state estimate. The specific formula is expressed as:

[0015]

[0016] in, K stands for the estimation of the system state at time k based on the observed ECU data at time k, and K stands for the a posteriori state estimation, which means the best guess of the system state after receiving and processing all the observed ECU data before time k. k It is expressed as a Kalman gain matrix, which is calculated based on the ratio of the prediction error covariance and the observation error covariance and is used to balance the predicted state and observed ECU data to obtain the best estimate, Z k It is represented as the observed ECU data containing noise obtained at time k. Secondly, by fusing the ECU data of different sensors, the ECU data is optimized and integrated to improve the robustness and accuracy of the system. The specific formula for ECU data optimization and integration is expressed as:

[0017]

[0018] Among them, X opt Represents the optimized and integrated ECU data, X i Represented as the ECU data of the i-th sensor, w i It is expressed as the weight of the i-th sensor ECU data, reflecting the importance of the i-th sensor ECU data in the integration process. j It is represented by the ECU data of the remaining sensors except the ECU data of the i-th sensor, and λ is represented by the adjustment parameter, which is used to control the influence of the optimization function on the final integration result. It is represented as an optimization function, which is used to process paired ECU data from different sensors. i and j represent the index of the sensor type serial number, and n represents the total number of sensor type serial numbers. By integrating the data from various sensors, the advantages of multi-modal ECU data fusion are used to intelligently integrate to improve the performance of the vehicle monitoring system. Then, after obtaining the optimized integrated data, decision-level fusion is performed to achieve the final judgment of the vehicle status. The specific formula is expressed as:

[0019] D=g(X opt )

[0020] Then, D represents the final decision output, g represents the decision function, based on the optimized integrated data X opt Information recognition is performed, and the fused data is used for actual vehicle monitoring decisions to improve the system's response speed and accuracy. Finally, according to the system's output feedback, the algorithm is adjusted and optimized to adapt to the ever-changing monitoring environment, ensuring that the system can continue to provide accurate monitoring data and adapt to different operating conditions. The specific formula is expressed as follows;

[0021] A new =A old +ΔA

[0022] Among them, A new Expressed as the updated state transfer matrix, A old It is represented as the state transfer matrix before updating, ΔA is represented as the adjustment amount of the state transfer matrix calculated according to the system feedback. The data optimization integration algorithm based on multimodal data fusion enables the system to effectively integrate multimodal data, optimize the data integration process, and improve the accuracy and robustness of vehicle monitoring.

[0023] Preferably, the intelligent data integration module includes a data intelligent management unit, which analyzes and manages the optimized and integrated data to ensure that the system can effectively process massive data and provide efficient and accurate analysis results for subsequent accident detection and supervision operations.

[0024] Preferably, the real-time accident detection module proposes a real-time traffic accident recognition algorithm based on abnormal motion detection to automatically identify traffic accidents and save on-site videos in real time to ensure that accident evidence can be recorded as soon as possible. Through efficient accident recognition technology and stable video storage functions, it ensures a rapid response in the event of an emergency and ensures the integrity and security of accident data.

[0025] Specifically, the real-time traffic accident recognition algorithm based on abnormal motion detection is as follows: First, a global foreground model is constructed to detect abnormal motion. By using the video data in the ECU data to identify abnormal motion related to traffic accidents, moving foreground objects and temporarily stopped objects can be detected. The global foreground model is constructed by the Gaussian mixture modeling method to estimate whether each pixel in the video data belongs to the foreground or the background. The specific formula is expressed as follows:

[0026]

[0027] Among them, P(x|M f ) is represented as the foreground category set M f The probability that the pixel x belongs to the foreground, K represents the number of Gaussian distributions in the foreground model, k represents the number index of Gaussian distributions in the foreground model, and Wk is represented as the weight of the kth Gaussian distribution, x is represented as the feature vector of the pixel block, μ k Represented as the mean vector of the kth Gaussian distribution, M f is represented as a foreground category set, f is represented as the index of the frame data in the video sequence, N(x|μ k ) is expressed as the mean μ k The multivariate Gaussian distribution centered on describes the shape of the k-th Gaussian distribution in the feature space. The specific formula is expressed as:

[0028]

[0029] Where d represents the dimension of the feature vector x, Σ k is represented as the covariance matrix of the kth Gaussian distribution, |·| is represented as the determinant operation, exp is represented as the exponential function, which is used to calculate the exponential term of the Gaussian density function, and T is represented as the transpose operation. The constructed Gaussian mixture model constitutes the basis of the global foreground modeling method, which is used to detect and identify foreground objects in video analysis, including abnormal motion related to traffic accidents. Secondly, the boundary of the traffic flow is determined by summing the foreground masks of all frames and dividing them by the total number of frames to identify vehicles that have deviated from the normal driving path and had an accident. The cumulative foreground mask obtained from the constructed global foreground model is used to estimate the boundary of the road. The specific formula is expressed as:

[0030]

[0031] Among them, r c It is represented as a cumulative traffic area map, which represents the average intensity of the traffic areas detected in all frames, r c is a binary image, where pixels with a value of 1 represent traffic areas, and pixels with a value of 0 represent non-traffic areas. F represents the total number of frames in the video sequence, which is used to estimate the total number of frames in the traffic area. m f The foreground mask of the fth frame is represented as a binary image, where pixels with a value of 1 represent foreground objects, and pixels with a value of 0 represent background objects. Then, by accumulating the traffic area map r c The threshold τ is applied to construct a binary traffic area map. The area above the threshold is marked as a traffic area, and the area below the threshold is marked as a non-traffic area. The specific formula is expressed as:

[0032]

[0033] Among them, r fIt is represented as the final traffic area binary map, which is a binary image, where pixels with a value of 1 represent belonging to the traffic area, and pixels with a value of 0 represent not belonging to the traffic area. τ represents the threshold calculated by the automatic image threshold segmentation technology. The boundary of the traffic area is automatically detected by the constructed cumulative traffic area map and the final traffic area binary map to identify vehicles that have deviated from the normal driving path and have accidents. Secondly, the moving direction of the vehicle in the traffic flow is determined by estimating the traffic direction. The traffic direction is estimated by analyzing the movement of the vehicle in continuous frames. The specific formula for calculating the vehicle speed vector component is expressed as:

[0034] v z =z efront -z elater ,v y =y efront -y elater

[0035] Among them, v z Expressed as the velocity component in the horizontal direction, v y Expressed as the velocity component in the vertical direction, z efront It is expressed as the mean z coordinate of the vehicle center of mass in the closer half of the frame sequence considered, z elater It is expressed as the mean z coordinate of the center of mass of the vehicle in the farther half of the frame sequence considered, y efront It is expressed as the mean y coordinate of the vehicle center of mass in the closer half of the frame sequence considered, y elater It is represented as the mean y coordinate of the vehicle center of mass in the farther half of the frame sequence considered, elater is represented as the mean of the vehicle center of mass position in the farther half of the frame sequence considered, and efront is represented as the mean of the vehicle center of mass position in the closer half of the frame sequence considered. The specific formula for vehicle direction estimation is expressed as:

[0036]

[0037] Among them, d i is represented as the estimated direction of vehicle i, representing the moving direction of the vehicle relative to the horizontal axis, arctan is represented as the inverse tangent function operation, and finally, the information obtained from the previous steps is used to build a first-order logic decision system to detect traffic accidents in real time. The specific formula is expressed as:

[0038]

[0039] Among them, U represents the set of all vehicles, u represents the index of the number of vehicles, Vehicle represents an actual tracked vehicle, Fast(u) represents that the speed of vehicle u is higher than the average speed of the traffic unit, Swerve(v) represents the accident sign information that the movement direction of vehicle u differs from the average direction of the traffic unit by more than 45 degrees, ShortDistance(u) represents the accident sign information that the movement safety distance of vehicle u is small, ξ represents the weight coefficient for avoiding false alarms caused by small movements, Rapid(u) represents the accident sign information that vehicle u has made an unexpected rapid lateral movement, ∧ represents the logical AND operation, and the system can effectively reduce the occurrence of traffic accidents and respond quickly when an accident occurs by performing abnormal vehicle motion detection and real-time traffic accident recognition based on abnormal motion detection.

[0040] Preferably, the Bluetooth communication module implements individual voice reminders for the driver through a Bluetooth headset, ensuring that the driver can receive alarm and prompt information in a timely manner. At the same time, the reminder information is transmitted to the driver through Bluetooth technology, reducing the impact on passengers and ensuring that the quiet environment for passengers in the vehicle is not disturbed.

[0041] Preferably, the remote supervision module transmits the vehicle's driving status and accident information to the supervision platform in real time through the high-precision positioning and stable communication functions of the Beidou system, ensuring that the supervision platform can instantly obtain the vehicle's operating information and accident details, providing reliable protection for remote monitoring, and enhancing the integrity and transparency of vehicle data during transmission and storage through the decentralization and immutability of blockchain technology, preventing data from being tampered with or lost, and automatically executing preset operations through smart contracts. The distributed architecture of blockchain enables vehicle data to be securely shared among supervision platforms, insurance companies, and maintenance agencies, thereby improving collaborative management efficiency and providing a reliable protection and traceability mechanism for remote monitoring.

[0042] Preferably, the driver feedback module adds an interactive feedback function to the vehicle terminal, so that the system can promptly confirm the receipt of alarm information, ensure that the alarm information is accurately received and processed, and enhance the user experience and operational convenience of the system.

[0043] Compared with the prior art, the present invention has the following beneficial effects:

[0044] 1. ECU data integration unit proposes a data optimization integration algorithm based on multimodal data fusion. First, by establishing an ECU data fusion model, the system can centrally integrate raw ECU data from multiple sensors, including engine, transmission system, vehicle dynamics, body control and environmental perception data. This centralized architecture ensures that all ECU data are processed on a unified platform, making data integration and optimization more efficient. Then, the Kalman filter is used to estimate and predict the system state, reducing the impact of sensor noise on the system and greatly improving the accuracy of the data. At the same time, the dynamic prediction model of the Kalman filter can update the system state based on historical data and new observation data, realizing accurate monitoring of the real-time state of the vehicle. Secondly, the ECU data In the process of data optimization and integration, a weighted integration method of multimodal data is adopted, and the importance and relevance of various sensor data are taken into consideration. Paired sensor data are processed by optimizing functions, thereby improving the accuracy of data integration and the robustness of the system. Finally, after the optimization and integration of multimodal data, the system can judge the vehicle status at the decision-making level, significantly improving the response speed and accuracy of the monitoring system. At the same time, the feedback-based algorithm optimization mechanism ensures that the system can maintain efficient performance in a constantly changing monitoring environment and adapt to different operating conditions. In summary, the data optimization and integration algorithm based on multimodal data fusion improves the accuracy, real-time and robustness of the entire vehicle monitoring system, and can provide reliable data support for vehicle status monitoring and decision-making.

[0045] 2. Real-time accident detection module proposes a real-time traffic accident recognition algorithm based on abnormal motion detection. First, the system constructs a global foreground model to effectively detect and identify abnormal motion related to traffic accidents. The model is based on Gaussian mixture modeling and can distinguish between foreground and background in video data, accurately locate moving and temporarily stopped vehicles, and thus identify potential accident risks. By calculating the multivariate Gaussian distribution of pixels of foreground objects, the system can automatically detect abnormal motion in complex traffic scenes. Secondly, the system uses the cumulative foreground mask of traffic flow to identify vehicles that have deviated from the normal driving path and may potentially cause an accident. By analyzing the vehicle's motion trajectory in consecutive frames, the system can accurately estimate the vehicle's moving direction and monitor whether it has sudden turns. The algorithm effectively defines the boundaries of traffic flow by constructing a binary traffic area map, and combines automatic image threshold segmentation technology to enhance the accuracy of traffic accident identification. Finally, the algorithm uses a first-order logic decision-making system to integrate the vehicle's speed, direction change and safety distance information to make accident warning decisions in real time. By incorporating the vehicle's speed, direction deviation and short-distance behavior into logical reasoning, the system can respond in time when the vehicle quickly moves sideways and deviates from the normal path, reducing the possibility of accidents and providing real-time feedback for accident management. In summary, the real-time traffic accident identification algorithm based on abnormal motion detection enables the system to not only enhance the intelligence of vehicle monitoring, but also improve the overall level of traffic safety management through efficient accident identification capabilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] The invention is further illustrated by the accompanying drawings, but the embodiments in the accompanying drawings do not constitute any limitation to the invention. A person skilled in the art can obtain other drawings based on the following drawings without creative work.

[0047] Figure 1 It is a structural schematic diagram of the present invention. DETAILED DESCRIPTION

[0048] 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.

[0049] See also Figure 1The present invention provides a vehicle monitoring system based on ECU data integration and intelligent management, including an ECU data reading module, an intelligent data integration module, a Bluetooth communication module, a real-time accident detection module, a remote supervision module, and a driver feedback module, characterized in that: the ECU data reading module is used to obtain the vehicle's operating status data in real time, including engine data, transmission system data, vehicle dynamic data, body control data, and environmental perception data, to ensure that the system can continuously monitor and provide accurate vehicle operating information; the intelligent data integration module includes an ECU data integration unit and a data intelligent management unit; the ECU data integration unit proposes a data optimization integration algorithm based on multimodal data fusion for intelligently integrating multimodal data information read from the ECU, and provides comprehensive vehicle operating status information by optimizing and integrating data from different sensors; the data intelligent management unit is used to It is used to intelligently manage the integrated ECU data, analyze the vehicle's maintenance needs and driving behavior information, and ensure the long-term and efficient operation of the vehicle. The Bluetooth communication module is used to realize the driver's individual voice reminders, and provide vehicle status and accident warning information in a timely manner to ensure that the driver focuses on driving and obtains necessary real-time information. The real-time accident detection module proposes a real-time traffic accident recognition algorithm based on abnormal motion detection to automatically identify the occurrence of traffic accidents. When an accident is detected, the camera is automatically triggered and the on-site video is saved to facilitate subsequent analysis and responsibility definition. The remote supervision module is used to transmit the vehicle's operating data to the supervision platform in real time through the Beidou system to ensure the safety and compliance of the vehicle. The driver feedback module is used for interactive feedback between the system and the driver. By providing operation suggestions, real-time data reports and post-accident processing solutions, it helps the driver to efficiently understand the vehicle's status and improve driving safety.

[0050] See also Figure 1 ,Furthermore, the ECU data reading module acquires the vehicle’s operating status data in real time through the vehicle’s hardware interface, including speed, engine status, and fuel consumption data, ensuring that the system can fully grasp the vehicle’s operating status, provide real-time vehicle information, and support subsequent data processing and monitoring functions.

[0051] See also Figure 1 Furthermore, the intelligent data integration module includes an ECU data integration unit. The ECU data integration unit proposes a data optimization integration algorithm based on multimodal data fusion for intelligently integrating multimodal data information read from the ECU, ensuring that different types of data can be processed uniformly and providing comprehensive vehicle operation status information.

[0052] See also Figure 1,Furthermore, the data optimization and integration algorithm based on multimodal data fusion is specifically as follows: First, an ECU data fusion model is established. Through the centralized ECU data architecture, the original ECU data from different sensors can be integrated, including engine ECU data, transmission system ECU data, vehicle dynamics ECU data, body control ECU data and environmental perception ECU data, to ensure that all sensor ECU data can be processed on a unified platform for subsequent ECU data optimization and integration. The specific formula is expressed as follows:

[0053] Y=HX+V

[0054] Among them, Y represents the fused observed ECU data, H represents the observation matrix, which represents the mapping of different sensor ECU data to the observation space, X represents the system state vector, which contains the state information of all sensors, and V represents the noise matrix. Secondly, the Kalman filter is used for state estimation and prediction to optimize and integrate ECU data from different sensors. The specific formula of the constructed dynamic prediction model is expressed as follows:

[0055]

[0056] in, It is represented as the predicted estimate of the system state at time k based on the information at time k-1. It represents the prior state estimate, which means the best guess of the system state before receiving the observed ECU data at time k. A is represented as the state transfer matrix, which represents the matrix of the evolution of the system state over time. The estimation of the system state at time k-1 based on the information at time k-1 represents the a posteriori state estimation, which represents the best guess of the system state after receiving the observed ECU data at time k. B is represented as the control input matrix. The control input U k Associated with state transfer, U k It is represented as the control input information applied at time k, W is represented as the process noise matrix, which represents the random variables in the state transfer process. W is Gaussian distributed and has zero mean and covariance matrix Q. k is represented as the discrete time index. By building a dynamic model to reduce the impact of sensor noise and improve the accuracy of ECU data, the constructed dynamic prediction model is then updated to use the new observed ECU data to correct the previously predicted state estimate. The specific formula is expressed as:

[0057]

[0058] in, K stands for the estimation of the system state at time k based on the observed ECU data at time k, and K stands for the a posteriori state estimation, which means the best guess of the system state after receiving and processing all the observed ECU data before time k. k It is expressed as a Kalman gain matrix, which is calculated based on the ratio of the prediction error covariance and the observation error covariance and is used to balance the predicted state and observed ECU data to obtain the best estimate, Z k It is represented as the observed ECU data containing noise obtained at time k. Secondly, by fusing the ECU data of different sensors, the ECU data is optimized and integrated to improve the robustness and accuracy of the system. The specific formula for ECU data optimization and integration is expressed as:

[0059]

[0060] Among them, X opt Represents the optimized and integrated ECU data, X i Represented as the ECU data of the i-th sensor, w i It is expressed as the weight of the i-th sensor ECU data, reflecting the importance of the i-th sensor ECU data in the integration process. j It is represented by the ECU data of the remaining sensors except the ECU data of the i-th sensor, and λ is represented by the adjustment parameter, which is used to control the influence of the optimization function on the final integration result. It is represented as an optimization function, which is used to process paired ECU data from different sensors. i and j represent the index of the sensor type serial number, and n represents the total number of sensor type serial numbers. By integrating the data from various sensors, the advantages of multi-modal ECU data fusion are used to intelligently integrate to improve the performance of the vehicle monitoring system. Then, after obtaining the optimized integrated data, decision-level fusion is performed to achieve the final judgment of the vehicle status. The specific formula is expressed as:

[0061] D=g(X opt )

[0062] Then, D represents the final decision output, g represents the decision function, based on the optimized integrated data X opt Information recognition is performed, and the fused data is used for actual vehicle monitoring decisions to improve the system's response speed and accuracy. Finally, according to the system's output feedback, the algorithm is adjusted and optimized to adapt to the ever-changing monitoring environment, ensuring that the system can continue to provide accurate monitoring data and adapt to different operating conditions. The specific formula is expressed as follows;

[0063] A new =A old +ΔA

[0064] Among them, Anew Expressed as the updated state transfer matrix, A old It is represented as the state transfer matrix before updating, ΔA is represented as the adjustment amount of the state transfer matrix calculated according to the system feedback. The data optimization integration algorithm based on multimodal data fusion enables the system to effectively integrate multimodal data, optimize the data integration process, and improve the accuracy and robustness of vehicle monitoring.

[0065] See also Figure 1 ,Furthermore, the intelligent data integration module includes a data intelligent management unit, which ,analyzes and manages the optimized and integrated data to ensure that the ,system can effectively process massive data and provide efficient and accurate ,analysis results for subsequent accident detection and supervision operations.

[0066] See also Figure 1 Furthermore, the real-time accident detection module proposes a real-time traffic accident recognition algorithm based on abnormal motion detection to automatically identify traffic accidents and save on-site videos in real time to ensure that accident evidence can be recorded in the first place. Through efficient accident recognition technology and stable video storage functions, it ensures rapid response in the event of an emergency and ensures the integrity and security of accident data.

[0067] See also Figure 1 ,Furthermore, the real-time traffic accident recognition algorithm based on abnormal motion detection is specifically as follows: First, a global foreground model is constructed to detect abnormal motion. By using the video data in the ECU data to identify abnormal motion related to traffic accidents, moving foreground objects and temporarily stopped objects can be detected. The global foreground model is constructed by the Gaussian mixture modeling method to estimate whether each pixel in the video data belongs to the foreground or the background. The specific formula is expressed as:

[0068]

[0069] Among them, P(x|M f ) is represented as the foreground category set M f The probability that the pixel x belongs to the foreground, K represents the number of Gaussian distributions in the foreground model, k represents the number index of Gaussian distributions in the foreground model, and W k is represented as the weight of the kth Gaussian distribution, x is represented as the feature vector of the pixel block, μ k Represented as the mean vector of the kth Gaussian distribution, M f is represented as a foreground category set, f is represented as the index of the frame data in the video sequence, N(x|μ k ) is expressed as the mean μ k The multivariate Gaussian distribution centered on describes the shape of the k-th Gaussian distribution in the feature space. The specific formula is expressed as:

[0070]

[0071] Where d represents the dimension of the feature vector x, Σ k is represented as the covariance matrix of the kth Gaussian distribution, |·| is represented as the determinant operation, exp is represented as the exponential function, which is used to calculate the exponential term of the Gaussian density function, and T is represented as the transpose operation. The constructed Gaussian mixture model constitutes the basis of the global foreground modeling method, which is used to detect and identify foreground objects in video analysis, including abnormal motion related to traffic accidents. Secondly, the boundary of the traffic flow is determined by summing the foreground masks of all frames and dividing them by the total number of frames to identify vehicles that have deviated from the normal driving path and had an accident. The cumulative foreground mask obtained from the constructed global foreground model is used to estimate the boundary of the road. The specific formula is expressed as:

[0072]

[0073] Among them, r c It is represented as a cumulative traffic area map, which represents the average intensity of the traffic areas detected in all frames, r c is a binary image, where pixels with a value of 1 represent traffic areas, and pixels with a value of 0 represent non-traffic areas. F represents the total number of frames in the video sequence, which is used to estimate the total number of frames in the traffic area. m f The foreground mask of the fth frame is represented as a binary image, where pixels with a value of 1 represent foreground objects, and pixels with a value of 0 represent background objects. Then, by accumulating the traffic area map r c The threshold τ is applied to construct a binary traffic area map. The area above the threshold is marked as a traffic area, and the area below the threshold is marked as a non-traffic area. The specific formula is expressed as:

[0074]

[0075] Among them, r f It is represented as the final traffic area binary map, which is a binary image, where pixels with a value of 1 represent belonging to the traffic area, and pixels with a value of 0 represent not belonging to the traffic area. τ represents the threshold calculated by the automatic image threshold segmentation technology. The boundary of the traffic area is automatically detected by the constructed cumulative traffic area map and the final traffic area binary map to identify vehicles that have deviated from the normal driving path and have accidents. Secondly, the moving direction of the vehicle in the traffic flow is determined by estimating the traffic direction. The traffic direction is estimated by analyzing the movement of the vehicle in continuous frames. The specific formula for calculating the vehicle speed vector component is expressed as:

[0076] v z =z efront -z elater,v y =y efront -y elater

[0077] Among them, v z Expressed as the velocity component in the horizontal direction, v y Expressed as the velocity component in the vertical direction, z efront It is expressed as the mean z coordinate of the vehicle center of mass in the closer half of the frame sequence considered, z elater It is expressed as the mean z coordinate of the center of mass of the vehicle in the farther half of the frame sequence considered, y efront It is expressed as the mean y coordinate of the vehicle center of mass in the closer half of the frame sequence considered, y elater It is represented as the mean y coordinate of the vehicle center of mass in the farther half of the frame sequence considered, elater is represented as the mean of the vehicle center of mass position in the farther half of the frame sequence considered, and efront is represented as the mean of the vehicle center of mass position in the closer half of the frame sequence considered. The specific formula for vehicle direction estimation is expressed as:

[0078]

[0079] Among them, d i is represented as the estimated direction of vehicle i, representing the moving direction of the vehicle relative to the horizontal axis, arctan is represented as the inverse tangent function operation, and finally, the information obtained from the previous steps is used to build a first-order logic decision system to detect traffic accidents in real time. The specific formula is expressed as:

[0080]

[0081] Among them, U represents the set of all vehicles, u represents the index of the number of vehicles, Vehicle represents an actual tracked vehicle, Fast(u) represents that the speed of vehicle u is higher than the average speed of the traffic unit, Swerve(v) represents the accident sign information that the movement direction of vehicle u differs from the average direction of the traffic unit by more than 45 degrees, ShortDistance(u) represents the accident sign information that the movement safety distance of vehicle u is small, ξ represents the weight coefficient for avoiding false alarms caused by small movements, Rapid(u) represents the accident sign information that vehicle u has made an unexpected rapid lateral movement, ∧ represents the logical AND operation, and the system can effectively reduce the occurrence of traffic accidents and respond quickly when an accident occurs by performing abnormal vehicle motion detection and real-time traffic accident recognition based on abnormal motion detection.

[0082] See also Figure 1Furthermore, the Bluetooth communication module realizes individual voice reminders for the driver through a Bluetooth headset, ensuring that the driver can receive alarm and prompt information in a timely manner. At the same time, the reminder information is transmitted to the driver through Bluetooth technology, reducing the impact on passengers and ensuring that the quiet environment for passengers in the vehicle is not disturbed.

[0083] See also Figure 1 Furthermore, the remote supervision module transmits the vehicle's driving status and accident information to the supervision platform in real time through the high-precision positioning and stable communication functions of the Beidou system, ensuring that the supervision platform can instantly obtain the vehicle's operating information and accident details, providing reliable protection for remote monitoring. The decentralization and immutability of blockchain technology enhance the integrity and transparency of vehicle data during transmission and storage, prevent data from being tampered with or lost, and automatically execute preset operations through smart contracts. The distributed architecture of blockchain enables vehicle data to be securely shared among supervision platforms, insurance companies, and maintenance agencies, improving collaborative management efficiency and providing a reliable guarantee and traceability mechanism for remote monitoring.

[0084] See also Figure 1 ,Furthermore, the driver feedback module adds an interactive feedback function to the ,vehicle terminal, so that the system can promptly confirm the receipt of the alarm ,information, ensuring that the alarm information is accurately received and processed, ,enhancing the user experience and operation convenience of the system.

[0085] When used specifically, first, the ECU data reading module is used to obtain the vehicle's operating status data in real time, including engine data, transmission system data, vehicle dynamic data, body control data and environmental perception data, to ensure that the system can continuously monitor and provide accurate vehicle operation information. Secondly, the intelligent data integration module includes the ECU data integration unit and the data intelligent management unit. The ECU data integration unit proposes a data optimization integration algorithm based on multimodal data fusion for intelligent integration of multimodal data information read from the ECU. By optimizing and integrating data from different sensors, comprehensive vehicle operating status information is provided. The data intelligent management unit is used to intelligently manage the integrated ECU data, analyze the vehicle's maintenance needs and driving behavior information, and ensure the long-term operation of the vehicle. Efficient operation, then, the Bluetooth communication module is used to realize the driver's individual voice reminder, provide vehicle status and accident warning information in time, ensure that the driver focuses on driving and obtains necessary real-time information, secondly, the accident real-time detection module proposes a real-time traffic accident recognition algorithm based on abnormal motion detection for automatic identification of traffic accidents, finally, when an accident is detected, the camera is automatically triggered and the on-site video is saved to facilitate subsequent analysis and responsibility definition, the remote supervision module is used to transmit the vehicle's operation data to the supervision platform in real time through the Beidou system to ensure the safety and compliance of the vehicle, and the driver feedback module is used for interactive feedback between the system and the driver, by providing operation suggestions, real-time data reports and post-accident processing solutions, to help drivers efficiently understand the status of the vehicle and improve driving safety.

[0086] Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments, or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A vehicle monitoring system based on ECU data integration and intelligent management, including an ECU data reading module, an intelligent data integration module, a Bluetooth communication module, a real-time accident detection module, a remote supervision module, and a driver feedback module, characterized in that: The ECU data reading module is used to obtain the vehicle's operating status data in real time, including engine data, transmission system data, vehicle dynamic data, body control data and environmental perception data, to ensure that the system can continuously monitor and provide accurate vehicle operation information. The intelligent data integration module includes an ECU data integration unit and a data intelligent management unit. The ECU data integration unit proposes a data optimization integration algorithm based on multimodal data fusion for intelligently integrating multimodal data information read from the ECU, and provides comprehensive vehicle operating status information by optimizing and integrating data from different sensors. The data intelligent management unit is used to intelligently manage the integrated ECU data, analyze the vehicle's maintenance needs and driving behavior information, and ensure the long-term efficiency of the vehicle. The Bluetooth communication module is used to realize the driver's individual voice reminder, provide vehicle status and accident warning information in time, ensure that the driver focuses on driving and obtains necessary real-time information, the real-time accident detection module proposes a real-time traffic accident recognition algorithm based on abnormal motion detection to automatically identify the occurrence of traffic accidents, automatically trigger the camera and save the on-site video when an accident is detected, which is convenient for subsequent analysis and responsibility definition, the remote supervision module is used to transmit the vehicle's operation data to the supervision platform in real time through the Beidou system to ensure the safety and compliance of the vehicle operation, and the driver feedback module is used for interactive feedback between the system and the driver, by providing operation suggestions, real-time data reports and post-accident processing solutions, to help the driver efficiently understand the vehicle status and improve driving safety.

2. The vehicle monitoring system based on ECU data integration and intelligent management according to claim 1, characterized in that: The ECU data reading module obtains the vehicle's operating status data in real time through the vehicle's hardware interface, including speed, engine status, and fuel consumption data, ensuring that the system can fully grasp the vehicle's operating status, provide real-time vehicle information, and support subsequent data processing and monitoring functions.

3. The vehicle monitoring system based on ECU data integration and intelligent management according to claim 1, characterized in that: The intelligent data integration module includes an ECU data integration unit and a data intelligent management unit. The ECU data integration unit proposes a data optimization integration algorithm based on multimodal data fusion for intelligently integrating multimodal data information read from the ECU, ensuring that different types of data can be processed uniformly and providing comprehensive vehicle operation status information. The data intelligent management unit analyzes and manages the optimized and integrated data to ensure that the system can effectively process massive data and provide efficient and accurate analysis results for subsequent accident detection and supervision operations.

4. The vehicle monitoring system based on ECU data integration and intelligent management according to claim 3 is characterized by: First, an ECU data fusion model is established. The centralized ECU data architecture can integrate the original ECU data from different sensors, including engine ECU data, transmission system ECU data, vehicle dynamics ECU data, body control ECU data, and environmental perception ECU data, to ensure that all sensor ECU data can be processed on a unified platform for subsequent ECU data optimization and integration. The specific formula is expressed as follows: Y=HX+V Among them, Y represents the fused observed ECU data, H represents the observation matrix, which represents the mapping of different sensor ECU data to the observation space, X represents the system state vector, which contains the state information of all sensors, and V represents the noise matrix. Secondly, the Kalman filter is used for state estimation and prediction to optimize and integrate ECU data from different sensors. The specific formula of the constructed dynamic prediction model is expressed as follows: in, It is represented as the predicted estimate of the system state at time k based on the information at time k-1. It represents the prior state estimate, which means the best guess of the system state before receiving the observed ECU data at time k. A is represented as the state transfer matrix, which represents the matrix of the evolution of the system state over time. The estimation of the system state at time k-1 based on the information at time k-1 represents the a posteriori state estimation, which represents the best guess of the system state after receiving the observed ECU data at time k. B is represented as the control input matrix. The control input U k Associated with state transfer, U k It is represented as the control input information applied at time k, W is represented as the process noise matrix, which represents the random variables in the state transfer process. W is Gaussian distributed and has zero mean and covariance matrix Q. k is represented as the discrete time index. By building a dynamic model to reduce the impact of sensor noise and improve the accuracy of ECU data, the constructed dynamic prediction model is then updated to use the new observed ECU data to correct the previously predicted state estimate. The specific formula is expressed as: in, K stands for the estimation of the system state at time k based on the observed ECU data at time k, and K stands for the a posteriori state estimation, which means the best guess of the system state after receiving and processing all the observed ECU data before time k. k It is expressed as a Kalman gain matrix, which is calculated based on the ratio of the prediction error covariance and the observation error covariance and is used to balance the predicted state and observed ECU data to obtain the best estimate, Z k It is represented as the observed ECU data containing noise obtained at time k. Secondly, by fusing the ECU data of different sensors, the ECU data is optimized and integrated to improve the robustness and accuracy of the system. The specific formula for ECU data optimization and integration is expressed as: Among them, X opt Represents the optimized and integrated ECU data, X i Represented as the ECU data of the i-th sensor, w i It is expressed as the weight of the i-th sensor ECU data, reflecting the importance of the i-th sensor ECU data in the integration process. j It is represented by the ECU data of the remaining sensors except the ECU data of the i-th sensor, and λ is represented by the adjustment parameter, which is used to control the influence of the optimization function on the final integration result. It is represented as an optimization function, which is used to process paired ECU data from different sensors. i and j represent the index of the sensor type serial number, and n represents the total number of sensor type serial numbers. By integrating the data from various sensors, the advantages of multi-modal ECU data fusion are used for intelligent integration to improve the performance of the vehicle monitoring system. Then, after obtaining the optimized integrated data, decision-level fusion is performed to achieve the final judgment of the vehicle status. The specific formula is expressed as: D=g(X opt ) Then, D represents the final decision output, g represents the decision function, based on the optimized integrated data X opt Information recognition is performed, and the fused data is used for actual vehicle monitoring decisions to improve the system's response speed and accuracy. Finally, according to the system's output feedback, the algorithm is adjusted and optimized to adapt to the ever-changing monitoring environment, ensuring that the system can continue to provide accurate monitoring data and adapt to different operating conditions. The specific formula is expressed as follows; IN new = Yes old +ΔA Among them, A new Expressed as the updated state transfer matrix, A old It is represented as the state transfer matrix before updating, ΔA is represented as the adjustment amount of the state transfer matrix calculated according to the system feedback. The data optimization integration algorithm based on multimodal data fusion enables the system to effectively integrate multimodal data, optimize the data integration process, and improve the accuracy and robustness of vehicle monitoring.

5. The vehicle monitoring system based on ECU data integration and intelligent management according to claim 1 is characterized in that: The Bluetooth communication module realizes individual voice reminders for the driver through a Bluetooth headset, ensuring that the driver can receive alarm and prompt information in a timely manner. At the same time, the reminder information is transmitted to the driver through Bluetooth technology, reducing the impact on passengers and ensuring that the quiet environment for passengers in the vehicle is not disturbed.

6. The vehicle monitoring system based on ECU data integration and intelligent management according to claim 1 is characterized in that: The real-time accident detection module proposes a real-time traffic accident recognition algorithm based on abnormal motion detection to automatically identify traffic accidents and save on-site videos in real time to ensure that accident evidence can be recorded in the first place. Through efficient accident recognition technology and stable video storage functions, it ensures rapid response in the event of an emergency and ensures the integrity and security of accident data.

7. The vehicle monitoring system based on ECU data integration and intelligent management according to claim 6, characterized in that: First, a global foreground model is constructed to detect abnormal motion. By using the video data in the ECU data to identify abnormal motion related to traffic accidents, moving foreground objects and temporarily stopped objects can be detected. The global foreground model is constructed by the Gaussian mixture modeling method to estimate whether each pixel in the video data belongs to the foreground or the background. The specific formula is expressed as: Among them, P(x|M f ) is represented as the foreground category set M f The probability that the pixel x belongs to the foreground, K represents the number of Gaussian distributions in the foreground model, k represents the number index of Gaussian distributions in the foreground model, and W k is represented as the weight of the kth Gaussian distribution, x is represented as the feature vector of the pixel block, μ k Represented as the mean vector of the kth Gaussian distribution, M f is represented as a foreground category set, f is represented as the index of the frame data in the video sequence, N(x|μ k ) is expressed as the mean μ k The multivariate Gaussian distribution centered on describes the shape of the k-th Gaussian distribution in the feature space. The specific formula is expressed as: Where d represents the dimension of the feature vector x, Σ k is represented as the covariance matrix of the kth Gaussian distribution, |·| is represented as the determinant operation, exp is represented as the exponential function, which is used to calculate the exponential term of the Gaussian density function, and T is represented as the transpose operation. The constructed Gaussian mixture model constitutes the basis of the global foreground modeling method, which is used to detect and identify foreground objects in video analysis, including abnormal motion related to traffic accidents. Secondly, the boundary of the traffic flow is determined by summing the foreground masks of all frames and dividing them by the total number of frames to identify vehicles that have deviated from the normal driving path and had an accident. The cumulative foreground mask obtained from the constructed global foreground model is used to estimate the boundary of the road. The specific formula is expressed as: Among them, r c It is represented as a cumulative traffic area map, which represents the average intensity of the traffic areas detected in all frames, r c is a binary image, where pixels with a value of 1 represent traffic areas, and pixels with a value of 0 represent non-traffic areas. F represents the total number of frames in the video sequence, which is used to estimate the total number of frames in the traffic area. m f The foreground mask of the fth frame is represented as a binary image, where pixels with a value of 1 represent foreground objects, and pixels with a value of 0 represent background objects. Then, by accumulating the traffic area map r c The threshold τ is applied to construct a binary traffic area map. The area above the threshold is marked as a traffic area, and the area below the threshold is marked as a non-traffic area. The specific formula is expressed as: Among them, r f It is represented as the final traffic area binary map, which is a binary image, where pixels with a value of 1 represent belonging to the traffic area, and pixels with a value of 0 represent not belonging to the traffic area. τ represents the threshold calculated by the automatic image threshold segmentation technology. The boundary of the traffic area is automatically detected by the constructed cumulative traffic area map and the final traffic area binary map to identify vehicles that have deviated from the normal driving path and have accidents. Secondly, the moving direction of the vehicle in the traffic flow is determined by estimating the traffic direction. The traffic direction is estimated by analyzing the movement of the vehicle in continuous frames. The specific formula for calculating the vehicle speed vector component is expressed as: v z =z efront -z elater ,v y =y efront -y elater Among them, v z Expressed as the velocity component in the horizontal direction, v y Expressed as the velocity component in the vertical direction, z efront It is expressed as the mean z coordinate of the vehicle center of mass in the closer half of the frame sequence considered, z elater It is expressed as the mean z coordinate of the center of mass of the vehicle in the farther half of the frame sequence considered, y efront It is expressed as the mean y coordinate of the vehicle center of mass in the closer half of the frame sequence considered, y elater It is represented as the mean y coordinate of the vehicle center of mass in the farther half of the frame sequence considered, elater is represented as the mean of the vehicle center of mass position in the farther half of the frame sequence considered, and efront is represented as the mean of the vehicle center of mass position in the closer half of the frame sequence considered. The specific formula for vehicle direction estimation is expressed as: Among them, d i is represented as the estimated direction of vehicle i, representing the moving direction of the vehicle relative to the horizontal axis, arctan is represented as the inverse tangent function operation, and finally, the information obtained from the previous steps is used to build a first-order logic decision system to detect traffic accidents in real time. The specific formula is expressed as: Among them, U represents the set of all vehicles, u represents the index of the number of vehicles, Vehicle represents an actual tracked vehicle, Fast(u) represents that the speed of vehicle u is higher than the average speed of the traffic unit, Swerve(v) represents the accident sign information that the movement direction of vehicle u differs from the average direction of the traffic unit by more than 45 degrees, ShortDistance(u) represents the accident sign information that the movement safety distance of vehicle u is small, ξ represents the weight coefficient for avoiding false alarms caused by small movements, Rapid(u) represents the accident sign information that vehicle u has made an unexpected rapid lateral movement, ∧ represents the logical AND operation, and the system can effectively reduce the occurrence of traffic accidents and respond quickly when an accident occurs by performing abnormal vehicle motion detection and real-time traffic accident recognition based on abnormal motion detection.

8. The vehicle monitoring system based on ECU data integration and intelligent management according to claim 1 is characterized in that: The remote monitoring module uses the high-precision positioning and stable communication functions of the Beidou system to transmit the vehicle's driving status and accident information to the monitoring platform in real time, ensuring that the monitoring platform can instantly obtain the vehicle's operating information and accident details, providing reliable protection for remote monitoring, and enhancing the integrity and transparency of vehicle data during transmission and storage through the decentralization and immutability of blockchain technology, preventing data from being tampered with or lost, and automatically executing preset operations through smart contracts. The distributed architecture of blockchain enables vehicle data to be securely shared among the monitoring platform, insurance companies, and maintenance agencies, improving collaborative management efficiency and providing a reliable protection and traceability mechanism for remote monitoring.

9. The vehicle monitoring system based on ECU data integration and intelligent management according to claim 1, characterized in that: The driver feedback module adds an interactive feedback function to the vehicle terminal, so that the system can promptly confirm the receipt of alarm information, ensure that the alarm information is accurately received and processed, and enhance the user experience and operational convenience of the system.

Images