Emergency rescue linkage method and system based on driving behavior analysis of vehicle state

Abstract

The application discloses an emergency rescue linkage method and system based on driving behavior analysis of vehicle state, relates to the field of big data processing, and comprises the following steps: collecting driving behavior, vehicle operation and environmental data in real time, constructing a standardized data set, training a normal driving behavior benchmark model by using a machine learning algorithm, and setting a characteristic threshold range; after comparing real-time data with the model, determining an abnormal level by a deviation value, correlatively analyzing an abnormal source in combination with vehicle operation and environmental data, evaluating a safety risk, and judging whether emergency rescue is needed; if rescue is needed, integrating data and uploading to a cloud platform, and according to vehicle position, abnormal level and surrounding resource distribution, linking rescue agencies, medical treatment, traffic control and insurance companies; and in the rescue process, the platform tracks rescue progress in real time.The application has the advantages that multi-dimensional data is integrated to realize accurate identification and source analysis of driving abnormalities, the response efficiency and accuracy of emergency rescue are greatly improved, and driving safety is ensured.

Description

Technical Field

[0001] This invention relates to the field of big data processing, and in particular to an emergency rescue linkage method and system based on driving behavior analysis of vehicle status. Background Technology

[0002] In recent years, by collecting real-time vehicle CAN bus data and driver behavior data, combined with machine learning algorithms, abnormal driving behavior or vehicle malfunctions can be dynamically identified, triggering automatic warnings and linking with emergency response centers. Currently, some high-end models are equipped with such systems, but they generally face challenges such as insufficient data fusion accuracy, high false alarm rates, and low efficiency in multi-source information collaboration. In the future, with the maturity of 5G, edge computing, and V2X technologies, the real-time performance and reliability of driving behavior analysis and emergency rescue will be further improved, driving a decrease in traffic accident fatalities and the intelligentization of emergency rescue systems.

[0003] Current vehicle emergency rescue methods on the market have significant disadvantages, primarily in their insufficient multi-dimensional collaboration and intelligence. They largely rely on single vehicle operation data or manual alarm triggers, lacking in-depth analysis and integration of dynamic and static characteristics of driving behavior, making it difficult to accurately identify the root cause of anomalies and prone to misjudgments or omissions. Simultaneously, information silos exist between departments, and the lack of a unified cloud platform for dispatching leads to delayed rescue responses and poor multi-party collaboration. Positioning is easily affected by complex terrain, and the rescue process lacks dynamic tracking and solution optimization mechanisms. Furthermore, they have poor adaptability to special scenarios and insufficient professional capabilities to handle special situations such as new energy vehicle malfunctions, making it difficult to form a precise, end-to-end control system from anomaly warning to rescue, thus limiting rescue efficiency and safety. Summary of the Invention

[0004] To improve existing methods and systems, this method provides an emergency rescue linkage method and system based on driving behavior analysis of vehicle status. This method integrates multi-dimensional data to achieve accurate identification and root cause analysis of driving anomalies, relies on a cloud platform to link multiple departments for collaborative rescue and dynamically optimize the plan, forming a closed-loop management of the entire process, which greatly improves the efficiency and accuracy of emergency rescue response and ensures the safety of drivers and passengers.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] Emergency rescue coordination methods based on vehicle status analysis of driving behavior include:

[0007] Based on the real-time collection of driving behavior data, basic vehicle operation data and environmental data by the sensor devices on the vehicle, a standardized dataset is obtained after cleaning, noise reduction and standardization.

[0008] Based on the obtained standardized driving behavior dataset, static and dynamic features of driving behavior are extracted. Based on normal driving behavior sample data, machine learning algorithms are used to train and construct a benchmark model of normal driving behavior. The feature threshold range of normal driving behavior is obtained through iterative training.

[0009] The real-time extracted driving behavior features are input into the normal driving behavior benchmark model. The deviation between the real-time driving behavior features and the normal driving behavior features in the benchmark model is calculated. The abnormal driving behavior is judged by setting a deviation threshold and the abnormal state level is marked.

[0010] Based on abnormal driving behavior data, abnormal vehicle operation data, and environmental data, the root cause type of abnormal vehicle status is determined through correlation analysis. Based on the root cause type and abnormal status level, a risk assessment matrix is ​​used to assess the vehicle's safety risk value and to make emergency rescue linkage judgments.

[0011] Based on the need for emergency rescue coordination, data related to abnormal vehicle status are integrated, transmitted to the cloud platform through in-vehicle terminal equipment, and the information is verified.

[0012] Based on the vehicle's current location, abnormal status level, and distribution of surrounding rescue resources, the cloud platform sends linkage data to rescue entities, medical institutions, traffic management departments, and insurance companies through linkage interfaces;

[0013] During emergency rescue operations, the cloud platform receives real-time location information of rescue vehicles, feedback from rescue personnel, and vehicle status update data uploaded by onboard terminal devices, enabling real-time tracking of the rescue progress.

[0014] Preferably, the process of acquiring standardized datasets by real-time collection of driving behavior data, basic vehicle operation data, and environmental data using on-vehicle sensor devices, followed by cleaning, noise reduction, and standardization, specifically includes:

[0015] The driving behavior data includes the driver's steering operation data, accelerator pedal travel data, brake pedal trigger frequency and force data, gear shifting data, and steering wheel rotation angle and angular velocity data.

[0016] The vehicle's basic operating data includes engine speed data, vehicle speed data, braking system pressure data, tire pressure data, engine oil pressure data, coolant temperature data, and battery voltage data.

[0017] The environmental data includes the vehicle's current driving location data, road condition level data of the driving section, real-time weather data, and light intensity data;

[0018] The collected data is cleaned, denoised, and standardized to obtain standardized driving behavior datasets, vehicle operation datasets, and environmental datasets.

[0019] Preferably, the step of extracting static and dynamic features of driving behavior based on the acquired standardized driving behavior dataset, training and constructing a benchmark model of normal driving behavior using machine learning algorithms based on normal driving behavior sample data, and obtaining the feature threshold range of normal driving behavior through iterative training specifically includes:

[0020] When extracting static features, a time window is set, and the average steering angle, average accelerator pedal travel and average braking frequency within the window are calculated. The number of gear shifts per unit mileage is recorded.

[0021] During dynamic feature extraction, the instantaneous change rate of steering angle, the rate of rise and fall of accelerator pedal travel, and the difference between peak and valley values ​​of braking force are calculated in real time. The continuous driving time and the number of fluctuations in driving behavior characteristics per unit time are also counted to form a feature set.

[0022] Based on normal driving behavior samples of vehicles of the same model, a normal driving behavior benchmark model is constructed using the gradient boosting tree algorithm. The extracted static and dynamic features are used as input vectors for multi-round iterative training and optimization to obtain the normal threshold range of each driving behavior feature.

[0023] Preferably, the step of inputting the real-time extracted driving behavior features into the normal driving behavior benchmark model, calculating the deviation value between the real-time driving behavior features and the normal driving behavior features in the benchmark model, and determining abnormal driving behavior by setting a deviation threshold and marking the abnormal state level specifically includes:

[0024] The static and dynamic features of driving behavior extracted in real time are input into the normal driving behavior benchmark model. The deviation values ​​between the real-time features and the normal feature benchmark values ​​in the model are calculated dimension by dimension, and the deviation data sequence corresponding to each feature is generated.

[0025] For deviation data sequences with different characteristics, differentiated deviation thresholds are set based on vehicle model and driving scenario, and anomaly alarms are triggered when the threshold is exceeded.

[0026] By combining the vehicle operation dataset, the initially identified abnormal driving behavior is further verified, and the abnormality level is marked for vehicles that are determined to be in an abnormal state.

[0027] Preferably, the step of determining the root cause type of vehicle abnormality through correlation analysis based on abnormal driving behavior data, abnormal vehicle operation data, and environmental data, assessing the vehicle's safety risk value using a risk assessment matrix based on the root cause type and abnormality level, and making emergency rescue linkage judgments specifically includes:

[0028] Based on abnormal driving behavior data, abnormal vehicle operation data, and environmental data, the three types of data are aligned according to the time sequence of abnormal occurrence. By using feature matching, data dimensions that are strongly correlated with the abnormal state are selected to determine the root cause type.

[0029] The root causes include driver error, vehicle component failure, and external environmental interference.

[0030] A risk assessment matrix is ​​constructed based on the root cause type, abnormal state level, and current driving environment. The risk weight of each dimension is quantified, and the safety risk value is calculated comprehensively. When the risk value exceeds the preset threshold, the emergency rescue linkage mechanism is triggered.

[0031] Preferably, the process of integrating vehicle abnormality data based on the need for emergency rescue coordination, transmitting it to the cloud platform via in-vehicle terminal equipment, and performing information verification specifically includes:

[0032] Once the emergency rescue linkage mechanism is activated, relevant data on the abnormal vehicle status are integrated, including the vehicle's unique identification information, real-time location information, abnormal status level, abnormal root cause location results, key vehicle operation data, driver's basic information, and real-time environmental data.

[0033] The integrated rescue information is fully encrypted using a symmetric encryption algorithm and transmitted to the cloud platform. The cloud platform receives and parses the encrypted information, and verifies the integrity and accuracy of the data through a verification code.

[0034] If there are missing fields or data errors, a targeted retransmission request is sent to the vehicle terminal. After the retransmission is completed and the verification is successful, a receipt confirmation is generated.

[0035] Preferably, the cloud platform sends linkage data to the rescue entity, medical institutions, traffic management departments, and insurance companies through a linkage interface based on the vehicle's current location, abnormal status level, and distribution of surrounding rescue resources. Specifically, this includes:

[0036] Rescue instructions are sent to the rescue entities through the cloud platform's linkage interface. These entities include nearby roadside assistance organizations, medical institutions, traffic management departments, and insurance companies.

[0037] Send vehicle location, abnormal status, and faulty component information to roadside assistance agencies, and dispatch rescue vehicles to the scene;

[0038] Send driver warning information and vehicle location data to medical institutions;

[0039] Send vehicle anomaly information and route information to traffic management departments to facilitate traffic control;

[0040] Send accident warning information to insurance companies;

[0041] The vehicle-mounted terminal device sends rescue progress information to the driver.

[0042] Preferably, during the emergency rescue process, the cloud platform receives real-time location information of rescue vehicles, feedback information from rescue personnel, and vehicle status update data uploaded by onboard terminal devices to track the rescue progress in real time, specifically including:

[0043] During emergency rescue operations, the cloud platform receives real-time location information of rescue vehicles, feedback from rescue personnel, and vehicle status updates to track the progress of the rescue in real time.

[0044] If there is congestion on the rescue route, worsening of vehicle malfunctions, or changes in the driver's injury, the cloud platform will adjust the rescue plan and reschedule the corresponding rescue entity.

[0045] Once the rescue vehicles arrive at the scene and complete the rescue operation, the cloud platform records all data from the entire rescue process, generates a rescue summary, and sends a rescue completion notification to all participating rescue entities, thus terminating the emergency rescue coordination process.

[0046] Furthermore, an emergency rescue linkage system based on driving behavior analysis of vehicle status is proposed, including:

[0047] Data acquisition and preprocessing module: Collects driving behavior, vehicle operation and environmental data in real time through on-board sensors, performs cleaning, noise reduction and standardization processing to generate structured datasets;

[0048] Feature extraction and model training module: Extracts static and dynamic features of driving behavior, uses machine learning algorithms to build a benchmark model of normal driving behavior, and determines the feature threshold range;

[0049] Anomaly detection and rating module: Calculates the deviation between driving behavior characteristics and the baseline model in real time, determines abnormal behavior based on thresholds, and rates the abnormal state.

[0050] Root cause analysis and risk assessment module: correlates driving behavior, vehicle operation and environmental data, identifies the types of abnormal root causes, and determines whether to trigger emergency rescue through a risk assessment matrix;

[0051] Data integration and transmission module: Encrypts and integrates abnormal vehicle data and transmits it to the cloud platform, verifying data integrity;

[0052] Rescue dispatch and coordination module: Based on vehicle location, anomaly level, and distribution of rescue resources, it sends coordination instructions to rescue organizations, medical institutions, etc., to coordinate rescue operations;

[0053] Rescue tracking and feedback module: Real-time monitoring of rescue progress, dynamic adjustment of rescue plan, recording of rescue results and termination of linkage process;

[0054] Processor: The processor is used to handle the calculation process of each formula and the construction calculation process of each model.

[0055] Compared with the prior art, the advantages of the present invention are:

[0056] By integrating multi-dimensional data on driving behavior, vehicle operation, and the environment, and through standardized preprocessing and precise extraction of dynamic and static features, this method leverages a machine learning benchmark model to achieve early and accurate identification and grading of abnormal behavior. Its core advantage lies in clarifying the root cause of anomalies through correlation analysis and scientifically determining the necessity of rescue by combining a risk assessment matrix, thus avoiding false triggers and missed rescues. Simultaneously, encrypted transmission and data verification ensure information security and reliability, while a cloud platform enables multi-departmental collaborative scheduling, real-time tracking of rescue progress, and dynamic optimization of the plan. This method achieves closed-loop management of the entire process from anomaly warning and risk assessment to rescue implementation, significantly improving the efficiency and accuracy of emergency rescue response and maximizing the safety of drivers and passengers. Attached Figure Description

[0057] Figure 1 This is a schematic diagram of the method proposed in this invention;

[0058] Figure 2 This is a schematic diagram of the data acquisition and preprocessing proposed in this invention;

[0059] Figure 3 This is a schematic diagram illustrating the feature extraction and normal behavior model construction proposed in this invention;

[0060] Figure 4 This is a schematic diagram of vehicle abnormal state recognition proposed in this invention;

[0061] Figure 5 This is a schematic diagram illustrating the location of the root cause of abnormal states proposed in this invention;

[0062] Figure 6 This is a schematic diagram of the emergency rescue information integration proposed in this invention;

[0063] Figure 7 This is a schematic diagram of the emergency rescue linkage proposed in this invention;

[0064] Figure 8 This is a schematic diagram illustrating the rescue process tracking and adjustment proposed in this invention. Detailed Implementation

[0065] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0066] An emergency rescue linkage system based on driving behavior analysis of vehicle status includes:

[0067] Data acquisition and preprocessing module: Collects driving behavior, vehicle operation and environmental data in real time through on-board sensors, performs cleaning, noise reduction and standardization processing to generate structured datasets;

[0068] Feature extraction and model training module: Extracts static and dynamic features of driving behavior, uses machine learning algorithms to build a benchmark model of normal driving behavior, and determines the feature threshold range;

[0069] Anomaly detection and rating module: Calculates the deviation between driving behavior characteristics and the baseline model in real time, determines abnormal behavior based on thresholds, and rates the abnormal state.

[0070] Root cause analysis and risk assessment module: correlates driving behavior, vehicle operation and environmental data, identifies the types of abnormal root causes, and determines whether to trigger emergency rescue through a risk assessment matrix;

[0071] Data integration and transmission module: Encrypts and integrates abnormal vehicle data and transmits it to the cloud platform, verifying data integrity;

[0072] Rescue dispatch and coordination module: Based on vehicle location, anomaly level, and distribution of rescue resources, it sends coordination instructions to rescue organizations, medical institutions, etc., to coordinate rescue operations;

[0073] Rescue tracking and feedback module: Real-time monitoring of rescue progress, dynamic adjustment of rescue plan, recording of rescue results and termination of linkage process;

[0074] Processor: The processor is used to handle the calculation process of each formula and the construction calculation process of each model.

[0075] See Figure 1 As shown, the emergency rescue linkage method based on vehicle status analysis of driving behavior includes:

[0076] Step 1: Real-time collection of driving behavior data, basic vehicle operation data, and environmental data based on vehicle-mounted sensor devices. After cleaning, noise reduction, and standardization, a standardized dataset is obtained.

[0077] Step 2: Based on the obtained standardized driving behavior dataset, extract the static and dynamic features of driving behavior. Based on the normal driving behavior sample data, use machine learning algorithms to train and build a benchmark model of normal driving behavior. Obtain the feature threshold range of normal driving behavior through iterative training.

[0078] Step 3: Input the real-time extracted driving behavior features into the normal driving behavior benchmark model, calculate the deviation value between the real-time driving behavior features and the normal driving behavior features in the benchmark model, determine the abnormal driving behavior by setting the deviation threshold, and mark the abnormal state level.

[0079] Step 4: Based on abnormal driving behavior data, abnormal vehicle operation data, and environmental data, determine the root cause type of abnormal vehicle status through correlation analysis. Based on the root cause type and abnormal status level, use a risk assessment matrix to assess the vehicle's safety risk value and make emergency rescue linkage judgments.

[0080] Step 5: Based on the need for emergency rescue coordination, integrate the relevant data on abnormal vehicle status, transmit it to the cloud platform through the vehicle terminal equipment, and verify the information;

[0081] Step Six: Based on the vehicle's current location, abnormal status level, and distribution of surrounding rescue resources, the cloud platform sends linkage data to the rescue entity, medical institutions, traffic management departments, and insurance companies through the linkage interface;

[0082] Step 7: During the emergency rescue operation, the cloud platform receives real-time location information of rescue vehicles, feedback information from rescue personnel, and vehicle status update data uploaded by on-board terminal equipment, and tracks the rescue progress in real time.

[0083] See Figure 2 As shown, based on the real-time collection of driving behavior data, basic vehicle operation data, and environmental data by sensor devices on the vehicle, after cleaning, noise reduction, and standardization processing, a standardized dataset is obtained, specifically including:

[0084] The driving behavior data includes the driver's steering operation data, accelerator pedal travel data, brake pedal trigger frequency and force data, gear shifting data, and steering wheel rotation angle and angular velocity data.

[0085] The vehicle's basic operating data includes engine speed data, vehicle speed data, braking system pressure data, tire pressure data, engine oil pressure data, coolant temperature data, and battery voltage data.

[0086] The environmental data includes the vehicle's current driving location data, road condition level data of the driving section, real-time weather data, and light intensity data;

[0087] The collected data is cleaned, denoised, and standardized to obtain standardized driving behavior datasets, vehicle operation datasets, and environmental datasets.

[0088] See Figure 3 As shown, based on the obtained standardized driving behavior dataset, static and dynamic features of driving behavior are extracted. Based on normal driving behavior sample data, a benchmark model of normal driving behavior is trained using machine learning algorithms. The feature threshold range of normal driving behavior is obtained through iterative training, specifically including:

[0089] When extracting static features, a time window is set, and the average steering angle, average accelerator pedal travel and average braking frequency within the window are calculated. The number of gear shifts per unit mileage is recorded.

[0090] During dynamic feature extraction, the instantaneous change rate of steering angle, the rate of rise and fall of accelerator pedal travel, and the difference between peak and valley values ​​of braking force are calculated in real time. The continuous driving time and the number of fluctuations in driving behavior characteristics per unit time are also counted to form a feature set.

[0091] Based on normal driving behavior samples of vehicles of the same model, a normal driving behavior benchmark model is constructed using the gradient boosting tree algorithm. The extracted static and dynamic features are used as input vectors for multi-round iterative training and optimization to obtain the normal threshold range of each driving behavior feature.

[0092] Specifically, based on the standardized driving behavior dataset, the time window segmentation method is used to divide the data into segments. The window duration is set to 60 seconds and the sliding step size is 10 seconds to ensure the timeliness and continuity of feature extraction. The data in each window is used as a set of independent feature extraction units to avoid the impact of fluctuations in a single data point on the accuracy of features.

[0093] In terms of static feature extraction, within each time window, the average steering angle (arithmetic mean of all steering angle data within the statistical window), average accelerator pedal travel (ratio of cumulative accelerator pedal travel within the calculated window to window duration), and average braking frequency (ratio of the number of brake pedal triggers within the statistical window to window duration) are calculated per unit time. At the same time, the collaborative features of accelerator and brake operations are extracted (the percentage of times the brake is triggered within 1 second after the accelerator pedal is released within the same window).

[0094] In terms of dynamic feature extraction, the steering angle change rate and accelerator pedal travel change rate are calculated by the feature difference between adjacent time windows. The braking force change amplitude is obtained by the difference between the maximum and minimum values ​​of braking force within the window. Combined with the timing module of the vehicle terminal, the continuous driving time is statistically recorded, and the trend change data of driving behavior characteristics during continuous driving are recorded simultaneously.

[0095] See Figure 4 As shown, the real-time extracted driving behavior features are input into the normal driving behavior benchmark model. The deviation between the real-time driving behavior features and the normal driving behavior features in the benchmark model is calculated. By setting a deviation threshold, abnormal driving behavior is judged, and the abnormal state level is marked. Specifically, this includes:

[0096] The static and dynamic features of driving behavior extracted in real time are input into the normal driving behavior benchmark model. The deviation values ​​between the real-time features and the normal feature benchmark values ​​in the model are calculated dimension by dimension, and the deviation data sequence corresponding to each feature is generated.

[0097] For deviation data sequences with different characteristics, differentiated deviation thresholds are set based on vehicle model and driving scenario, and anomaly alarms are triggered when the threshold is exceeded.

[0098] By combining the vehicle operation dataset, the initially identified abnormal driving behavior is further verified, and the abnormality level is marked for vehicles that are determined to be in an abnormal state.

[0099] Specifically, the core driving behavior features extracted in real time are organized according to a preset format and then input into the calibrated normal driving behavior benchmark model in batches. For each core feature, the deviation between the real-time feature value and the normal threshold range of the corresponding feature in the benchmark model is calculated. The deviation quantification adopts a two-dimensional evaluation of absolute deviation and relative deviation, and the duration of each feature deviation is recorded to avoid misjudging instantaneous fluctuations as abnormalities.

[0100] Differentiated deviation thresholds are set based on feature type. The relative deviation threshold for static features is set at 30%-40%, and the relative deviation threshold for dynamic features is set at 25%-35%. When the relative deviation of a single feature exceeds the corresponding threshold and lasts for ≥2 seconds, or when the relative deviations of three or more different types of features simultaneously exceed 70% of their respective thresholds and last for ≥1 second, a preliminary judgment of abnormal driving behavior is triggered, and the abnormal feature type and deviation degree are marked.

[0101] Standardized real-time vehicle operation data is used to construct a driving behavior-vehicle operation correlation analysis matrix. The temporal synchronization and causal relationship between the initially determined abnormal driving behavior and the vehicle operation data are analyzed. The focus is on verifying whether the following abnormal vehicle operation is accompanied by abnormal driving behavior: engine speed fluctuation exceeding 50% within 3 seconds, sudden increase or decrease of braking system pressure exceeding 20%, tire pressure drop exceeding 10% within 1 minute, oil pressure below the normal threshold, coolant temperature exceeding the warning value, etc. If abnormal driving behavior occurs simultaneously with one or more of the above abnormal vehicle operation, it is confirmed that the abnormal driving behavior is not due to intentional operation by the driver.

[0102] Based on the combined verification results of abnormal driving behavior and the vehicle operation data status, the final judgment of the abnormal vehicle status is completed: if there is only abnormal driving behavior but no abnormal vehicle operation, it is judged as a potential risk state of the vehicle caused by abnormal driving operation; if abnormal driving behavior and abnormal vehicle operation coexist, it is judged as a substantial abnormal vehicle status; then, according to the degree of impact on safety, it is divided into three levels of abnormality: mild abnormality, moderate abnormality, and severe abnormality.

[0103] See Figure 5 As shown, based on abnormal driving behavior data, abnormal vehicle operation data, and environmental data, the root cause type of abnormal vehicle status is determined through correlation analysis. Based on the root cause type and abnormal status level, a risk assessment matrix is ​​used to evaluate the vehicle's safety risk value, and emergency rescue coordination judgment is performed, specifically including:

[0104] Based on abnormal driving behavior data, abnormal vehicle operation data, and environmental data, the three types of data are aligned according to the time sequence of abnormal occurrence. By using feature matching, data dimensions that are strongly correlated with the abnormal state are selected to determine the root cause type.

[0105] The root causes include driver error, vehicle component failure, and external environmental interference.

[0106] A risk assessment matrix is ​​constructed based on the root cause type, abnormal state level, and current driving environment. The risk weight of each dimension is quantified, and the safety risk value is calculated comprehensively. When the risk value exceeds the preset threshold, the emergency rescue linkage mechanism is triggered.

[0107] Specifically, for the initially identified root cause types, precise positioning is carried out: Driver operation error type: Combining continuous driving time, trends in driving behavior characteristics, and driver physiological state data, such as fatigue information collected by vehicle cameras, the error type is determined, including fatigue driving, operational misjudgment, and improper emergency response. The severity of the error is assessed by the duration and scope of the erroneous action. Vehicle component failure type: Based on specific indicators of abnormal vehicle operation data, the faulty component is located. For example, abnormal braking system pressure corresponds to brake line or brake pad failure, and a sudden drop in tire pressure corresponds to tire damage or valve core failure. The degree of failure is determined by combining the historical operating data of the faulty component, such as cumulative usage time and past failure records. External environmental interference type: Combining information such as changes in road conditions, sudden weather changes, and sudden obstacles in environmental data, the interference type is identified. The duration and scope of the interference are assessed by the degree of impact on the vehicle's driving trajectory and driving operation, determining whether it is a temporary interference or a persistent risk.

[0108] See Figure 6 As shown, based on the need for emergency rescue coordination, relevant data on abnormal vehicle conditions are integrated, transmitted to the cloud platform via in-vehicle terminal equipment, and the information is verified. Specifically, this includes:

[0109] Once the emergency rescue linkage mechanism is activated, relevant data on the abnormal vehicle status are integrated, including the vehicle's unique identification information, real-time location information, abnormal status level, abnormal root cause location results, key vehicle operation data, driver's basic information, and real-time environmental data.

[0110] The integrated rescue information is fully encrypted using a symmetric encryption algorithm and transmitted to the cloud platform. The cloud platform receives and parses the encrypted information, and verifies the integrity and accuracy of the data through a verification code.

[0111] If there are missing fields or data errors, a targeted retransmission request is sent to the vehicle terminal. After the retransmission is completed and the verification is successful, a receipt confirmation is generated.

[0112] Specifically, based on risk assessment and various data collected and processed in the early stages, a standardized rescue information dataset is constructed and classified and integrated into three main categories: basic identification information, core anomaly information, and related supporting information. Basic identification information includes the vehicle's unique identification code, license plate number, vehicle model, year of manufacture, driver's name and contact information, and vehicle insurance information. Core anomaly information includes the anomaly status level, anomaly trigger time, anomaly root cause type and details, faulty component location results, and current safety risk level. Related supporting information includes the vehicle's real-time location coordinates, driving segment name and road type, real-time traffic and weather data, key abnormal parameters of vehicle operation, and abnormal driving behavior segments within 5 consecutive seconds. During the integration process, a preset data template is used for formatting to ensure that all information fields are complete and formatted uniformly, avoiding information redundancy or missing information.

[0113] See Figure 7 As shown, the cloud platform sends coordinated data to rescue entities, medical institutions, traffic management departments, and insurance companies through a linkage interface based on the vehicle's current location, abnormal status level, and distribution of surrounding rescue resources. Specifically, this includes:

[0114] The system sends rescue instructions to the rescue entities through the cloud platform's linkage interface. These entities include nearby roadside assistance organizations, medical institutions, traffic management departments, and insurance companies.

[0115] Send vehicle location, abnormal status and faulty parts information to roadside assistance agencies, and dispatch rescue vehicles to the scene;

[0116] Send driver warning information and vehicle location to medical institutions;

[0117] Send vehicle anomaly information and route information to traffic management departments to facilitate traffic control;

[0118] Send accident warning information to insurance companies;

[0119] The vehicle-mounted terminal device sends rescue progress information to the driver.

[0120] Specifically, a rescue plan decision-making model is constructed by combining the vehicle's abnormal status level, real-time location, distribution of surrounding rescue resources, and road condition information to comprehensively determine rescue priorities and resource allocation logic. For severe abnormalities, medical rescue and road clearing are given the highest priority, with priority given to dispatching the nearest ambulance and tow truck, and planning the optimal rescue route that avoids congested sections. For moderate abnormalities, road repair vehicles are prioritized for on-site inspection, and traffic management departments are simultaneously notified to prepare for traffic control in the surrounding area. For minor abnormalities, road rescue personnel are dispatched to remotely guide drivers on temporary handling, and nearby rescue vehicles are arranged to stand by if necessary. The plan clearly defines the task division, departure time, estimated arrival time, and coordination requirements for each rescue entity.

[0121] The cloud platform has opened real-time voice and text communication channels to enable information exchange among road rescue, medical, traffic management, and insurance entities. Dedicated personnel are assigned to monitor the progress of each rescue entity in real time. When conflicts arise in rescue routes or resource shortages, a conflict resolution mechanism is immediately activated: the routes of rescue vehicles are adjusted, and the order of arrival is optimized; backup rescue resources are dispatched from surrounding areas to supplement on-site forces; and traffic management departments are coordinated to ensure that rescue channels remain open.

[0122] See Figure 8 As shown, during the emergency rescue operation, the cloud platform receives real-time location information of rescue vehicles, feedback information from rescue personnel, and vehicle status update data uploaded by onboard terminal devices, enabling real-time tracking of the rescue progress. Specifically, this includes:

[0123] During emergency rescue operations, the cloud platform receives real-time location information of rescue vehicles, feedback from rescue personnel, and vehicle status updates to track the progress of the rescue in real time.

[0124] If there is congestion on the rescue route, worsening of vehicle malfunctions, or changes in the driver's condition, the cloud platform will adjust the rescue plan and reschedule the corresponding rescue entity.

[0125] Once the rescue vehicles arrive at the scene and complete the rescue operation, the cloud platform records all data from the entire rescue process, generates a rescue summary, and sends a rescue completion notification to all participating rescue entities, thus terminating the emergency rescue coordination process.

[0126] Specifically, based on the collected rescue time-series data, a progress monitoring dashboard is built to compare the real-time rescue progress with the planned time in the preset rescue plan, intuitively presenting the progress deviations of each rescue entity. A progress warning threshold is set: if the rescue vehicle's travel delay exceeds 10 minutes, the on-site handling time exceeds the estimated duration by 50%, or abnormal vehicle parameters continue to deteriorate, an early warning is immediately triggered, and a dedicated person in the coordination center intervenes immediately. Progress deviation information is synchronized to each rescue entity via voice or text channels, reminding them to adjust their action pace in a timely manner.

[0127] In response to any emergencies discovered during the tracking process, a differentiated dynamic adjustment mechanism is activated. If the rescue route is congested, the cloud platform uses real-time traffic data to replan the optimal route, pushes it to the rescue vehicle through the navigation system, and notifies the traffic management department to manage traffic at key nodes on the new route. If the vehicle's abnormal condition worsens, additional rescue resources are immediately allocated, medical rescue priority is simultaneously increased, and medical institutions are coordinated to reserve resources for critical care. If the driver's condition changes, emergency vehicles are dispatched to the scene in advance to guide on-site rescue personnel in implementing temporary first aid measures. The adjusted plan must be reissued to all relevant rescue entities, and the new rescue process is implemented after confirmation of receipt.

[0128] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0129] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0130] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An emergency rescue coordination method based on vehicle status analysis of driving behavior, characterized in that, include: Based on the real-time collection of driving behavior data, basic vehicle operation data and environmental data by the sensor devices on the vehicle, a standardized dataset is obtained after cleaning, noise reduction and standardization. Based on the obtained standardized driving behavior dataset, static and dynamic features of driving behavior are extracted. Based on normal driving behavior sample data, machine learning algorithms are used to train and construct a benchmark model of normal driving behavior. The feature threshold range of normal driving behavior is obtained through iterative training. The real-time extracted driving behavior features are input into the normal driving behavior benchmark model. The deviation between the real-time driving behavior features and the normal driving behavior features in the benchmark model is calculated. The abnormal driving behavior is judged by setting a deviation threshold and the abnormal state level is marked. Based on abnormal driving behavior data, abnormal vehicle operation data, and environmental data, the root cause type of abnormal vehicle status is determined through correlation analysis. Based on the root cause type and abnormal status level, a risk assessment matrix is ​​used to assess the vehicle's safety risk value and to make emergency rescue linkage judgments. Based on the need for emergency rescue coordination, data related to abnormal vehicle status are integrated, transmitted to the cloud platform through in-vehicle terminal equipment, and the information is verified. Based on the vehicle's current location, abnormal status level, and distribution of surrounding rescue resources, the cloud platform sends linkage data to rescue entities, medical institutions, traffic management departments, and insurance companies through linkage interfaces; During emergency rescue operations, the cloud platform receives real-time location information of rescue vehicles, feedback from rescue personnel, and vehicle status update data uploaded by onboard terminal devices, enabling real-time tracking of the rescue progress.

2. The emergency rescue linkage method based on vehicle status analysis according to claim 1, characterized in that, The vehicle-based sensor devices collect driving behavior data, basic vehicle operation data, and environmental data in real time. After cleaning, denoising, and standardization processing, a standardized dataset is obtained, specifically including: The driving behavior data includes the driver's steering operation data, accelerator pedal travel data, brake pedal trigger frequency and force data, gear shifting data, and steering wheel rotation angle and angular velocity data. The vehicle's basic operating data includes engine speed data, vehicle speed data, braking system pressure data, tire pressure data, engine oil pressure data, coolant temperature data, and battery voltage data. The environmental data includes the vehicle's current driving location data, road condition level data of the driving section, real-time weather data, and light intensity data; The collected data is cleaned, denoised, and standardized to obtain standardized driving behavior datasets, vehicle operation datasets, and environmental datasets.

3. The emergency rescue linkage method based on vehicle status analysis according to claim 1, characterized in that, Based on the acquired standardized driving behavior dataset, static and dynamic features of driving behavior are extracted. Based on normal driving behavior sample data, a benchmark model of normal driving behavior is trained using machine learning algorithms. The feature threshold range of normal driving behavior is obtained through iterative training, specifically including: When extracting static features, a time window is set, and the average steering angle, average accelerator pedal travel and average braking frequency within the window are calculated. The number of gear shifts per unit mileage is recorded. During dynamic feature extraction, the instantaneous change rate of steering angle, the rate of rise and fall of accelerator pedal travel, and the difference between peak and valley values ​​of braking force are calculated in real time. The continuous driving time and the number of fluctuations in driving behavior characteristics per unit time are also counted to form a feature set. Based on normal driving behavior samples of vehicles of the same model, a normal driving behavior benchmark model is constructed using the gradient boosting tree algorithm. The extracted static and dynamic features are used as input vectors for multi-round iterative training and optimization to obtain the normal threshold range of each driving behavior feature.

4. The emergency rescue linkage method based on vehicle status analysis according to claim 1, characterized in that, The process of inputting real-time extracted driving behavior features into a normal driving behavior benchmark model, calculating the deviation between the real-time driving behavior features and the normal driving behavior features in the benchmark model, and determining abnormal driving behavior by setting a deviation threshold and marking the abnormal state level specifically includes: The static and dynamic features of driving behavior extracted in real time are input into the normal driving behavior benchmark model. The deviation values ​​between the real-time features and the normal feature benchmark values ​​in the model are calculated dimension by dimension, and the deviation data sequence corresponding to each feature is generated. For deviation data sequences with different characteristics, differentiated deviation thresholds are set based on vehicle model and driving scenario, and anomaly alarms are triggered when the threshold is exceeded. By combining the vehicle operation dataset, the initially identified abnormal driving behavior is further verified, and the abnormality level is marked for vehicles that are determined to be in an abnormal state.

5. The emergency rescue linkage method based on vehicle status analysis according to claim 1, characterized in that, The process of determining the root cause type of vehicle abnormality through correlation analysis based on abnormal driving behavior data, abnormal vehicle operation data, and environmental data, and then using a risk assessment matrix to evaluate the vehicle's safety risk value based on the root cause type and abnormality level, and making emergency rescue coordination decisions specifically includes: Based on abnormal driving behavior data, abnormal vehicle operation data, and environmental data, the three types of data are aligned according to the time sequence of abnormal occurrence. By using feature matching, data dimensions that are strongly correlated with the abnormal state are selected to determine the root cause type. The root causes include driver error, vehicle component failure, and external environmental interference. A risk assessment matrix is ​​constructed based on the root cause type, abnormal state level, and current driving environment. The risk weight of each dimension is quantified, and the safety risk value is calculated comprehensively. When the risk value exceeds the preset threshold, the emergency rescue linkage mechanism is triggered.

6. The emergency rescue linkage method based on vehicle status analysis according to claim 1, characterized in that, The process of integrating vehicle abnormality data for emergency rescue coordination, transmitting it to the cloud platform via onboard terminal equipment, and verifying the information specifically includes: Once the emergency rescue linkage mechanism is activated, relevant data on the abnormal vehicle status are integrated, including the vehicle's unique identification information, real-time location information, abnormal status level, abnormal root cause location results, key vehicle operation data, driver's basic information, and real-time environmental data. The integrated rescue information is fully encrypted using a symmetric encryption algorithm and transmitted to the cloud platform. The cloud platform receives and parses the encrypted information, and verifies the integrity and accuracy of the data through a verification code. If there are missing fields or data errors, a targeted retransmission request is sent to the vehicle terminal. After the retransmission is completed and the verification is successful, a receipt confirmation is generated.

7. The emergency rescue linkage method based on vehicle status analysis according to claim 1, characterized in that, The cloud platform, based on the vehicle's current location, abnormal status level, and distribution of surrounding rescue resources, sends coordinated data to rescue entities, medical institutions, traffic management departments, and insurance companies via a linkage interface. Specifically, this includes: Rescue instructions are sent to the rescue entities through the cloud platform's linkage interface. These entities include nearby roadside assistance organizations, medical institutions, traffic management departments, and insurance companies. Send vehicle location, abnormal status, and faulty component information to roadside assistance agencies, and dispatch rescue vehicles to the scene; Send driver warning information and vehicle location data to medical institutions; Send vehicle anomaly information and route information to traffic management departments to facilitate traffic control; Send accident warning information to insurance companies; The vehicle-mounted terminal device sends rescue progress information to the driver.

8. The emergency rescue linkage method based on vehicle status analysis according to claim 1, characterized in that, During the emergency rescue operation, the cloud platform receives real-time location information of rescue vehicles, feedback from rescue personnel, and vehicle status update data uploaded by onboard terminal devices to track the rescue progress in real time. Specifically, this includes: During emergency rescue operations, the cloud platform receives real-time location information of rescue vehicles, feedback from rescue personnel, and vehicle status updates to track the progress of the rescue in real time. If there is congestion on the rescue route, worsening of vehicle malfunctions, or changes in the driver's injury, the cloud platform will adjust the rescue plan and reschedule the corresponding rescue entity. Once the rescue vehicles arrive at the scene and complete the rescue operation, the cloud platform records all data from the entire rescue process, generates a rescue summary, and sends a rescue completion notification to all participating rescue entities, thus terminating the emergency rescue coordination process.

9. An emergency rescue linkage system based on driving behavior analysis of vehicle status, used to implement the emergency rescue linkage method based on driving behavior analysis of vehicle status as described in any one of claims 1-8, characterized in that, include: Data acquisition and preprocessing module: Collects driving behavior, vehicle operation and environmental data in real time through on-board sensors, performs cleaning, noise reduction and standardization processing to generate structured datasets; Feature extraction and model training module: Extracts static and dynamic features of driving behavior, uses machine learning algorithms to build a benchmark model of normal driving behavior, and determines the feature threshold range; Anomaly detection and rating module: Calculates the deviation between driving behavior characteristics and the baseline model in real time, determines abnormal behavior based on thresholds, and rates the abnormal state. Root cause analysis and risk assessment module: correlates driving behavior, vehicle operation and environmental data, identifies the types of abnormal root causes, and determines whether to trigger emergency rescue through a risk assessment matrix; Data integration and transmission module: Encrypts and integrates abnormal vehicle data and transmits it to the cloud platform, verifying data integrity; Rescue dispatch and coordination module: Based on vehicle location, anomaly level, and distribution of rescue resources, it sends coordination instructions to rescue organizations, medical institutions, traffic management departments, and insurance companies to coordinate rescue operations; Rescue tracking and feedback module: Real-time monitoring of rescue progress, dynamic adjustment of rescue plan, recording of rescue results and termination of linkage process; Processor: The processor is used to handle the calculation process of each formula and the construction calculation process of each model.

Images