A vehicle-mounted safety data dynamic classification and multi-module recovery method and system

Abstract

The application provides a vehicle-mounted safety data dynamic classification and multi-module recovery method and system, including the following steps: during device operation, performing multi-level detection on safety data and generating a detection report containing data state; based on the detection report and real-time use records of the safety data, using a pre-trained machine learning model to dynamically classify the safety data, and dividing the safety data into at least vehicle machine binding data that needs to be repaired locally on the vehicle machine and cloud recovery data that can be recovered through cloud backup; and performing a differentiated recovery operation according to the classification result of the dynamic classification. The application solves the problem of how to implement multi-level, automated and deep detection of safety data, and based on the dynamic classification result, intelligently match the optimal recovery mode for different types of data, while integrating efficient cloud backup and intelligent decision mechanism, thereby comprehensively improving the reliability and recovery speed of data protection.

Description

Technical Field

[0001] This invention relates to the field of intelligent cockpit technology, and more specifically, to a method and system for dynamic classification and multi-module recovery of vehicle safety data. Background Technology

[0002] In the fields of smart cockpit and vehicle-to-everything (V2X) technology, secure data (such as encryption keys, digital certificates, and biometric information) is crucial for ensuring the secure and reliable startup and operation of the system. Current technologies primarily rely on simple local backup mechanisms and basic existence checks to protect this type of data. For example, during device startup, the system only verifies the existence of critical data, lacking in-depth analysis of data integrity, consistency, and timeliness. When data is corrupted or lost, current solutions generally rely on engineers' manual experience for troubleshooting, employing generic data recovery tools or replacing the entire device—a cumbersome and inefficient process. Furthermore, while some systems have introduced cloud backup functionality, the recovery process typically requires complex manual authorization and operation, failing to achieve rapid automation. Existing solutions cannot dynamically classify and manage data based on its importance and usage scenarios, nor do they offer one-click, precise recovery strategies for different data types. Consequently, the real-time performance, accuracy, and efficiency of data protection are insufficient to meet the high standards required by modern smart cockpit and V2X scenarios. Summary of the Invention

[0003] The problem addressed by this invention is how to achieve multi-level, automated deep detection of secure data, and intelligently match the optimal recovery mode for different types of data based on dynamic classification results, while integrating efficient cloud backup and intelligent decision-making mechanisms, thereby comprehensively improving the reliability and recovery speed of data protection.

[0004] To address the aforementioned issues, this invention provides a method for dynamic classification and multi-module recovery of in-vehicle safety data, comprising the following steps: During device operation, multi-level detection of safety data is performed, and a detection report containing data status is generated; based on the detection report and real-time usage records of the safety data, a pre-trained machine learning model is used to dynamically classify the safety data, dividing it into at least two categories: vehicle-bound data requiring local repair within the vehicle system and cloud-recoverable data recoverable via cloud backup; based on the classification results of the dynamic classification, differentiated recovery operations are performed; if the data is classified as vehicle-bound data, the local recovery mode is initiated, and the data is rewritten using a security data flashing tool; if the data is classified as cloud-recoverable data, the cloud recovery mode is automatically triggered, and, under the authorization of the main account, encrypted backup data is retrieved from the cloud secure storage area and refreshed to the vehicle system via incremental push.

[0005] The technical effects achieved by adopting this solution are as follows: Compared with existing technologies, the multi-layered detection mechanism can accurately locate problems such as data corruption and key mismatch, changing the traditional technology's crude approach of only judging "whether data is stored" and providing accurate diagnosis for subsequent intelligent recovery. The differentiated recovery strategy based on dynamic classification completely avoids a complete system wipe. For vehicle-mounted data, partial repair and incremental updates are used, greatly shortening recovery time and improving the success rate. For cloud-based data recovery, an unattended automated recovery process is achieved. Incremental pushes significantly reduce data traffic and recovery time, fundamentally solving the problems of low efficiency and poor user experience associated with manual recovery. Through machine learning models, data is classified and managed in real-time and dynamically, enabling the system to adapt to changes in data importance and usage scenarios. Based on system load, it intelligently decides recovery priorities and rationally allocates computing and network resources, ensuring rapid recovery of critical functions while guaranteeing the overall smoothness and stability of the vehicle-mounted system.

[0006] Furthermore, during device operation, multi-level checks are performed on security data, specifically including: consistency checks and timeliness checks. Consistency checks include verifying whether the encryption algorithm of the encrypted data is correct and whether the key matches; timeliness checks include checking whether the timestamped data is within the valid time range.

[0007] The technical effects achieved by adopting this solution are as follows: Consistency detection effectively identifies data logic errors caused by software upgrade mistakes, key rotation failures, or physical damage to the storage area, preventing the system from operating with potential faults and improving security from the source. Timeliness detection, targeting security data with lifecycles such as digital certificates and temporary tokens, proactively prevents system function interruptions caused by data expiration by verifying the validity of their timestamps. This ensures that the security data accessed by the system at any time is within its valid window, guaranteeing the continuity and compliance of business functions.

[0008] Furthermore, based on the detection report and real-time usage records of safety data, a pre-trained machine learning model is used to dynamically classify the safety data, dividing it into at least vehicle-mounted data that needs to be repaired locally on the vehicle and cloud-recoverable data that can be restored through cloud backup. The dynamic classification specifically includes the following criteria: the criticality of the data to device startup, the frequency of data usage, and the functional scenarios in which the data is applied.

[0009] The technical effects achieved by adopting this technical solution are as follows: Based on the dynamic classification of the above multi-dimensional standards, the system can prioritize the allocation of limited computing resources, storage bandwidth and recovery time windows to the most critical and frequently used data, thereby significantly improving the overall system resource utilization efficiency and the overall economy of the recovery process.

[0010] Furthermore, machine learning models are trained using decision trees or neural network algorithms and are periodically updated and optimized using new data samples.

[0011] The technical effects achieved by adopting this solution are as follows: Decision tree algorithms can accurately handle multi-dimensional classification rules, while neural networks excel at capturing complex nonlinear features. Both can deeply analyze the implicit correlations between the criticality of data, usage frequency, and functional scenarios, thereby achieving high-precision, multi-factor collaborative classification far exceeding that of manual rules, effectively avoiding mismatches in recovery strategies caused by misclassification. By regularly updating and optimizing the model with new data samples, the classification model can dynamically adapt to changes in user habits, software function iterations, and the emergence of new security threats. This mechanism ensures that the system's classification strategy does not become ineffective over time, maintaining its advanced and adaptable nature, and solving the inherent defects of static rule systems being rigid and difficult to maintain.

[0012] Furthermore, if the data is classified as vehicle-mounted data, the local recovery mode is activated, and the data is rewritten using a secure data flashing tool. This includes the following steps: When performing partial repair on vehicle-mounted data, the extent of data corruption is first analyzed, and only the damaged parts are repaired; for critical data, the remote external module upgrade mode is entered, and incremental upgrade packages are used for updates.

[0013] The technical effects achieved by adopting this solution are as follows: By first analyzing the extent of data damage, precise repair is performed only on the damaged portions, avoiding the numerous unnecessary operations caused by the overall data write-through in traditional solutions. This significantly reduces the amount of data manipulation, thereby shortening the recovery time from minutes to seconds, greatly improving repair efficiency and user experience. Partial repair reduces the number of write cycles to storage units and lowers the processor load during the recovery process, which helps extend hardware lifespan and ensures the smooth operation of other vehicle functions. For critical data, recovery is performed through a remote external module upgrade mode, a dedicated engineering channel with high privileges and high security. In this mode, incremental upgrade packages are used for updates, bypassing potential interference from the conventional operating system and enabling direct and controllable writing to secure storage areas. This ensures the high reliability and immutability of the core secure data recovery process, fundamentally eliminating new risks introduced by the recovery process itself.

[0014] Furthermore, if the data is classified as cloud-recovered data, the cloud recovery mode will be automatically triggered. Under the authorization of the main account, encrypted backup data will be retrieved from the secure cloud storage area and refreshed to the vehicle system in an incremental manner. Specifically, the triggering and execution of the cloud recovery mode does not require manual intervention and is automatically initiated by the system after detecting data anomalies.

[0015] The technical effects achieved by adopting this solution are as follows: From triggering and authorization to data acquisition and refreshing, the entire process does not require manual operation by maintenance personnel or users. This not only significantly reduces labor costs and the risk of human error, but also enables users to complete data repair without being aware of it, resulting in a revolutionary improvement in user experience.

[0016] Furthermore, cloud-based data recovery also includes the following steps: when the vehicle's infotainment system is powered off, the cloud-based data recovery is automatically encrypted and synchronized to the secure cloud storage area, and the backup version and timestamp are recorded.

[0017] The technical effects achieved by adopting this solution are as follows: Backups are automatically triggered at a fixed time when the vehicle's infotainment system is powered off. This avoids consuming system resources while the vehicle is running and ensures that data is archived in real time after each use, effectively preventing data loss due to sudden failures and guaranteeing the up-to-dateness and integrity of the backup data. Backup data is encrypted before transmission and storage, and combined with a secure cloud storage area, end-to-end data security protection is achieved, effectively resisting external attacks or theft risks and meeting compliance requirements for sensitive data such as user biometrics and digital certificates.

[0018] Furthermore, the method for dynamic classification and multi-module recovery of vehicle safety data also includes the following steps: based on the detection report and classification results, and combined with the current load of the central processing unit and memory, determine the recovery priority and timing of different safety data.

[0019] The technical effects achieved by adopting this solution are as follows: By sensing the load of the central processing unit and memory in real time, the system can autonomously decide when to execute recovery tasks. Under high load, non-critical recovery tasks are suspended to ensure the smooth operation of the vehicle's core functions; under low load, resources are concentrated to quickly complete the recovery, achieving seamless parallel operation of the recovery process and the user's normal driving experience, and resolving the conflict between recovery tasks and system performance competing for resources.

[0020] To address the aforementioned issues, this invention provides a multi-module recovery system for implementing the dynamic classification and multi-module recovery method for vehicle safety data provided by any of the above technical solutions. The multi-module recovery system includes: a detection module, a dynamic classification module, a multi-modal recovery module, a cloud backup and recovery management module, and an intelligent decision-making module. The detection module performs multi-level detection on safety data and generates a detection report; the dynamic classification module dynamically classifies safety data based on the detection report and the usage status of the safety data; the multi-modal recovery module calls the corresponding recovery modality to perform recovery operations according to the classification results; the cloud backup and recovery management module enables secure communication with the cloud, encrypted data backup, and recovery; and the intelligent decision-making module coordinates the various modules and determines the priority and timing of the recovery process based on the system status.

[0021] The technical effects achieved by adopting this solution are as follows: The system, through the closed-loop collaboration of five modules—detection, classification, decision-making, recovery, and backup—covers the entire chain from problem discovery and diagnostic analysis to repair execution. This completely changes the inefficient model that previously relied on manual sequential operations, achieving end-to-end automation of secure data operation and maintenance. The intelligent decision-making module, acting as the system's brain, coordinates all modules and performs intelligent scheduling based on global information. This enables the system to autonomously make optimal decisions, resolving efficiency bottlenecks and consistency issues caused by switching between multiple tools and manual judgment. The clear responsibilities and interfaces of each module enhance the system's maintainability and scalability.

[0022] Furthermore, the multimodal recovery module includes a local repair unit and a process control unit; the local repair unit is used to operate on the data bound to the vehicle system; the process control unit interacts with the cloud backup and recovery management module to achieve automated recovery of cloud-based data.

[0023] The technical effects achieved by adopting this solution are as follows: It separates the responsibilities of local and cloud recovery, adhering to the single responsibility principle, resulting in a clear system architecture and low coupling between modules. When an upgrade or modification to a particular recovery mode is required, it can be done only within the corresponding unit without altering the entire recovery module, greatly improving system maintainability and future functional expansion capabilities. The process control unit, as a dedicated interface with the cloud backup and recovery management module, encapsulates all automated triggering, authorization, data retrieval, and refresh steps. This centralized process management eliminates manual intervention points, ensuring seamless end-to-end connection and stable execution of the cloud recovery process, fundamentally preventing recovery failures caused by omitted steps or incorrect execution order, thereby improving the overall system's automation level and execution efficiency.

[0024] In summary, the technical solutions described above in this application have one or more of the following advantages or beneficial effects: i) By introducing a multi-layered detection mechanism that includes consistency and timeliness, the entire process of security data management is made intelligent and automated. From detection and classification to recovery, the system completes the process autonomously, greatly reducing manual intervention and improving operational efficiency and reliability. ii) Through differentiated recovery strategies, partial repair and incremental upgrades are used for vehicle-mounted data, and automatic triggering and incremental push are used for cloud-recovered data, achieving precision and efficiency in the recovery process and significantly shortening recovery time. iii) Through the separation of responsibilities and collaborative operation of the five functional modules of detection, classification, recovery, cloud management, and intelligent decision-making, an automated assurance system integrating diagnosis, decision-making, and execution is constructed, improving the maintainability and scalability of the system. Attached Figure Description

[0025] Figure 1This is a flowchart of a method for dynamic classification and multi-module recovery of vehicle safety data in an embodiment of the present invention;

[0026] Figure 2 This is a module diagram of a vehicle safety data dynamic classification and multi-module recovery system according to an embodiment of the present invention. Detailed Implementation

[0027] The purpose of this invention is to provide a method and system for dynamic classification and multi-module recovery of vehicle safety data, which can achieve intelligent diagnosis, accurate classification and efficient automated recovery of vehicle safety data as needed.

[0028] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0029] See Figures 1-2 This invention provides a method for dynamic classification and multi-module recovery of vehicle safety data, comprising the following steps:

[0030] S1: During device operation, perform multi-level detection on security data and generate a detection report that includes the data status;

[0031] S2: Based on the detection report and real-time usage records of safety data, the safety data is dynamically classified using a pre-trained machine learning model, and the safety data is divided into at least vehicle-mounted data that needs to be repaired locally on the vehicle and cloud-recoverable data that can be restored through cloud backup.

[0032] S3: Perform differential recovery operation based on the classification results of dynamic classification;

[0033] S31: If the data is classified as vehicle-mounted data, then start the local recovery mode and rewrite the data using the secure data flashing tool;

[0034] S32: If the data is classified as cloud recovery data, the cloud recovery mode will be automatically triggered. Under the authorization of the main account, encrypted backup data will be obtained from the secure cloud storage area and refreshed to the vehicle system in an incremental manner.

[0035] See Figure 1 During device operation, security data undergoes multi-level testing, specifically including consistency testing and timeliness testing. Consistency testing verifies whether the encryption algorithm of the encrypted data is correct and whether the key matches. Timeliness testing checks whether the timestamped data is within its valid time range.

[0036] Specifically, S1: When the in-vehicle intelligent cockpit device is started or the preset operating trigger conditions are met, the system begins to execute the security data protection process. First, the detection module performs multi-level detection on various security data stored in the device (such as vehicle private keys, digital certificates, biometric templates, etc.). This detection not only determines whether the data exists, but also deeply assesses its status, ultimately generating a structured detection report that clearly records the health status of each piece of data. S2: The dynamic classification module starts. The dynamic classification module receives the detection report from the detection module and real-time recorded security data usage records (such as access frequency and modification time). This module integrates a pre-trained machine learning model, which analyzes and calculates the input information and outputs dynamic classification results. The core criteria for classification are the security attributes and recovery characteristics of the data, which divides the secure data into at least two categories: the first category is vehicle-mounted data, which is highly sensitive and hardware-bound, and must and can only be repaired locally in the vehicle; the second category is cloud-recoverable data, which can be recovered through cloud backup mechanisms. S3: The system enters the recovery execution phase. Based on the above classification results, the system calls completely different recovery paths and performs differential recovery operations. S31: If the data is classified as vehicle-mounted data, the local recovery mode is activated, and the local secure data flashing tool is called to rewrite the faulty data; S32: If the data is classified as cloud-based recovery data, the cloud recovery mode is automatically triggered. After authorization by the main account, the encrypted backup data is obtained from the cloud secure storage area, and the data is refreshed to the target storage area in the vehicle-mounted system through incremental push technology.

[0037] See Figure 1 Based on the detection report and real-time usage records of safety data, a pre-trained machine learning model is used to dynamically classify the safety data, dividing it into at least vehicle-mounted data that needs to be repaired locally on the vehicle and cloud-recoverable data that can be restored through cloud backup. The dynamic classification is based on the criteria of the data’s criticality to device startup, the frequency of data usage, and the functional scenarios in which the data is applied.

[0038] Specifically, the criticality of device startup is determined by the model to assess whether the data is a necessary link in the system startup chain. The frequency of data usage is analyzed by the model to determine the number of times the data is accessed per unit of time. The functional scenarios in which the data is applied are identified by the model to determine which functions the data primarily serves. Based on a comprehensive analysis of these criteria, the model ultimately categorizes the data into either vehicle-mounted data or cloud-restored data.

[0039] Preferably, the classification results are updated in real time, and the classification strategy is automatically adjusted according to changes in the data to ensure classification accuracy and timeliness.

[0040] See Figure 1Machine learning models are trained using decision trees or neural network algorithms and are periodically updated and optimized using new data samples.

[0041] Specifically, firstly, a large amount of historical security data samples needs to be collected, and experts label each sample with its correct category according to classification criteria (such as vehicle-mounted data or cloud-recovered data). Then, these labeled samples are used to train a selected decision tree or neural network model, continuously adjusting the model's internal parameters through algorithms to enable it to learn classification rules. To prevent model aging and ensure its adaptability to new data patterns and user habits, the system periodically collects new, labeled data samples to incrementally train or retrain the deployed model, thereby updating and optimizing the model and maintaining its classification accuracy.

[0042] See Figure 1 If the data is classified as vehicle-mounted data, the local recovery mode is activated, and the data is rewritten using a secure data flashing tool. The specific steps include: when performing partial repair on vehicle-mounted data, the extent of data damage is first analyzed, and only the damaged parts are repaired; for critical data, the remote external module upgrade mode is entered, and incremental upgrade packages are used for updating.

[0043] Specifically, firstly, the system attempts to perform partial repair. The recovery tool precisely analyzes the damaged areas or extent of the data, and then repairs or rewrites only those damaged parts. Secondly, for specific critical data (such as underlying security data affecting system startup), when a deep flash is required, the system guides the device into a high-privilege remote external module upgrade mode (REM upgrade mode). In this mode, professionals can use a diagnostic interface to precisely update the target secure storage area using a pre-prepared incremental upgrade package containing only the differences, thereby further improving recovery efficiency while ensuring security.

[0044] See Figure 1 If the data is classified as cloud recovery data, the cloud recovery mode will be automatically triggered. Under the authorization of the main account, encrypted backup data will be obtained from the secure cloud storage area and refreshed to the vehicle system in an incremental way. Specifically, the triggering and execution of the cloud recovery mode does not require manual intervention and is automatically initiated by the system after detecting data anomalies.

[0045] Specifically, the entire cloud-based recovery process is triggered and executed without manual intervention. Once the detection module detects anomalies in the cloud-based recovery data and generates a report, the intelligent decision-making module automatically analyzes the data and triggers the recovery process. The system does not wait for instructions from maintenance personnel; professionals only need to authorize through the main account. The cloud backup and recovery management module retrieves secure data from the cloud and uses incremental push to refresh the data to the corresponding secure storage area of ​​the vehicle's infotainment system, reducing data transfer volume and recovery time.

[0046] See Figure 1 The cloud-based data recovery process also includes the following steps: when the vehicle's infotainment system is powered off, the cloud-based data recovery is automatically encrypted and synchronized to the secure cloud storage area, and the backup version and timestamp are recorded.

[0047] Specifically, each time the vehicle's infotainment system powers down normally, a backup task is automatically initiated. This task encrypts all data categorized as cloud-based recovery data and then synchronizes it to the secure cloud storage area via a secure network channel. Simultaneously, the cloud backup and recovery management module records detailed metadata for this backup, including a unique backup version number and a timestamp of backup creation, ensuring accurate selection and use of the corresponding backup version when recovery is needed.

[0048] See Figure 1 The method for dynamic classification and multi-module recovery of vehicle safety data also includes the following steps: based on the detection report and classification results, and combined with the current load of the central processing unit (CPU) and memory, determine the recovery priority and timing of different safety data.

[0049] Specifically, the intelligent decision-making module continuously monitors the system's resource status, especially the load on the central processing unit and memory. By combining comprehensive detection reports (the severity of data problems) and classification results (the criticality of data), it dynamically determines the recovery priority and timing for different types of secure data.

[0050] For example, when the system detects the loss of "vehicle-to-everything (V2X) data" that affects vehicle startup, and the current CPU and memory load is low, it will immediately set this data as the highest priority and attempt to restore it. Conversely, if the system is currently performing high-load calculations (such as navigation planning), and the lost data is the user configuration of the entertainment system (data restored from the cloud), the system may postpone this restoration task until system resources are available, thereby ensuring the smooth operation of the vehicle's core functions.

[0051] See Figures 1-2This invention provides a multi-module recovery system for implementing the vehicle safety data dynamic classification and multi-module recovery method provided by any of the above-mentioned technical solutions. The multi-module recovery system includes: a detection module, a dynamic classification module, a multi-modal recovery module, a cloud backup and recovery management module, and an intelligent decision-making module. The detection module is used to perform multi-level detection on safety data and generate a detection report; the dynamic classification module is used to dynamically classify safety data based on the detection report and the usage of safety data; the multi-modal recovery module is used to call the corresponding recovery mode to perform recovery operations according to the classification results; the cloud backup and recovery management module is used to realize secure communication with the cloud, encrypted data backup and recovery; and the intelligent decision-making module is used to coordinate the modules and decide the priority and timing of the recovery process according to the system status.

[0052] Specifically, the detection module is the system's sensing device, responsible for performing multi-level detection of security data and generating additional test reports containing detailed status information. The dynamic classification module is the system's analysis device, receiving detection reports and data usage records, and outputting dynamic classification results of the data through a built-in machine learning model. The multimodal recovery model is the system's execution device, invoking either local recovery or cloud recovery modes to complete the repair operation based on the classification results. The cloud backup and recovery management module is the system's "security bridge," responsible for managing all data interactions with the cloud, including encrypted uploading of backup data, decrypted downloading of recovery data, and version control. The intelligent decision-making module is the system's command center, connected to all other modules, responsible for coordinating workflows and deciding the priority and timing of recovery based on global information (data status, system load).

[0053] Furthermore, the multimodal recovery module includes a local repair unit and a process control unit; the local repair unit is used to operate on the data bound to the vehicle system; the process control unit interacts with the cloud backup and recovery management module to achieve automated recovery of cloud-based data.

[0054] Specifically, the local repair unit is responsible for operating on secure data categorized as vehicle-mounted data. This unit integrates dedicated data flashing tools and local repair algorithms to perform highly secure data recovery tasks locally on the vehicle's infotainment system. The process control unit manages the automated processes related to cloud recovery. It interacts with the cloud backup and recovery management module, controlling the entire sequence from triggering and authorization to data acquisition and push, ensuring efficient and automated data recovery from the cloud.

[0055] While the present invention has been disclosed above, it is not limited thereto. Any person skilled in the art can make various modifications and alterations without departing from the spirit and scope of the invention; therefore, the scope of protection of the present invention should be determined by the scope defined in the claims.

Claims

1. A method for dynamic classification and multi-module recovery of vehicle safety data, characterized in that, Includes the following steps: During equipment operation, multi-level detection of security data is performed, and a detection report containing data status is generated; The multi-level detection specifically includes: consistency detection and timeliness detection. The consistency detection includes verifying whether the encryption algorithm of the encrypted data is correct and whether the key matches. The timeliness detection includes detecting whether the timestamped data is within the valid time range. Based on the detection report and real-time usage records of the security data, a pre-trained machine learning model is used to dynamically classify the security data, dividing it into at least vehicle-mounted data that needs to be repaired locally on the vehicle and cloud-recoverable data that can be restored through cloud backup. Based on the classification results of the dynamic classification, perform a differentiated recovery operation; If the data is classified as vehicle-mounted data, the local recovery mode is activated, and the data is rewritten using the secure data flashing tool. If the data is classified as cloud recovery data, the cloud recovery mode will be automatically triggered. Under the authorization of the main account, encrypted backup data will be retrieved from the secure cloud storage area and refreshed to the vehicle system in an incremental manner.

2. The method for dynamic classification and multi-module recovery of vehicle safety data according to claim 1, characterized in that, Based on the detection report and real-time usage records of security data, a pre-trained machine learning model is used to dynamically classify the security data, dividing it into at least two categories: vehicle-bound data that needs to be repaired locally on the vehicle system and cloud-recoverable data that can be restored via cloud backup. Specifically, this includes the following steps: The dynamic classification module receives detection reports and real-time recorded security data usage records from the detection module. The dynamic classification module integrates a pre-trained machine learning model, which analyzes and calculates the input information and outputs dynamic classification results. The core criteria for classification are the security attributes and recovery characteristics of the data, which at least divides the security data into vehicle-mounted data and cloud-recoverable data. The dynamic classification specifically includes: The criteria for dynamic classification are the criticality of the data to device startup, the frequency of data usage, and the functional scenarios in which the data is applied.

3. The method for dynamic classification and multi-module recovery of vehicle safety data according to claim 2, characterized in that, The machine learning model is trained using decision tree or neural network algorithms and is periodically updated and optimized using new data samples.

4. The method for dynamic classification and multi-module recovery of vehicle safety data according to claim 1, characterized in that, If the data is classified as vehicle-mounted data, then the local recovery mode is activated, and the data is rewritten using a secure data flashing tool, specifically including the following steps: When performing partial repair on the vehicle-mounted data, the extent of data damage is first analyzed, and only the damaged parts are repaired. Specifically, for critical data, an incremental upgrade package is used to update it when entering the remote external module upgrade mode.

5. The method for dynamic classification and multi-module recovery of vehicle safety data according to claim 4, characterized in that, If the data is classified as cloud-recovered data, the cloud recovery mode is automatically triggered. Under the authorization of the main account, encrypted backup data is retrieved from the secure cloud storage area and refreshed to the vehicle's infotainment system in an incremental push manner. Specifically, this includes: The cloud-based recovery mode is triggered and executed automatically by the system after detecting data anomalies, without the need for manual intervention.

6. The method for dynamic classification and multi-module recovery of vehicle safety data according to claim 5, characterized in that, The cloud-based data recovery process also includes the following steps: When the vehicle's infotainment system is powered off, the cloud-based recovery data is automatically encrypted and synchronized to the cloud-based secure storage area, and the backup version and timestamp are recorded.

7. The method for dynamic classification and multi-module recovery of vehicle safety data according to claim 1, characterized in that, The method for dynamic classification and multi-module recovery of vehicle safety data also includes the following steps: Based on the detection report and classification results, and considering the current CPU and memory load, the recovery priority and timing for different types of security data are determined.

8. A multi-module recovery system, characterized in that, For implementing the vehicle safety data dynamic classification and multi-module recovery method according to any one of claims 1 to 7, the multi-module recovery system includes: The detection module is used to perform multi-level detection on the security data and generate the detection report; A dynamic classification module is used to dynamically classify the security data based on the detection report and the usage of the security data; A multimodal recovery module, which is used to call the corresponding recovery modality to perform a recovery operation based on the classification result; A cloud backup and recovery management module, which is used to achieve secure communication with the cloud, encrypted data backup and recovery; The intelligent decision-making module is used to coordinate the various modules and determine the priority and timing of the recovery process based on the system status.

9. The multi-module recovery system according to claim 8, characterized in that, The multimodal recovery module includes: A local repair unit, which is used to operate on the vehicle-mounted data; The process control unit interacts with the cloud backup and recovery management module to achieve automated recovery of the cloud-based data.

Images