A multi-source signal full-link processing system and method

Abstract

The application discloses a kind of multi-source signal full-link processing system and method, belong to intelligent network connection car and data processing technical field.The system includes multi-source signal acquisition module, multi-source data time synchronization and preprocessing module, event-driven acquisition control module, multi-source data fusion analysis module, structured storage module and data upload and cloud collaborative module.The application is associated with the same time road environment data and vehicle state data by the systematic collection of newly added driver behavior data, generates triad data unit, establishes complete causal chain, simultaneously improves data value density by adjusting acquisition strategy dynamically through event-driven mechanism, realizes the assetization management and value secondary amplification of data by structured storage and cloud collaborative.The application makes the value of raw data significantly improve due to the newly added dimension, and can be used in driving behavior analysis, road improvement, traffic management and automatic driving algorithm training and other fields.

Description

Technical Field

[0001] This invention belongs to the field of intelligent connected vehicle and data processing technology. It is a multi-source signal full-link processing system and method, specifically involving a human-vehicle-road multi-source signal full-link processing system and method for data value-added and driver behavior data fusion. Background Technology

[0002] With the widespread adoption of electric vehicles and energy storage power stations, the safe operation of battery packs has become paramount. With the rapid development of intelligent connected vehicle technology, modern vehicles are generally equipped with road environment perception devices such as forward-facing cameras, millimeter-wave radar, ultrasonic radar, GPS, and high-precision map modules, as well as vehicle status monitoring devices such as wheel speed sensors, inertial measurement units, and yaw rate sensors. These devices can collect large amounts of road environment and vehicle status data in real time and are widely used in the development and verification of ADAS, AEB, and ACC functions. However, although the data from the "vehicle" and "road" dimensions are collected and stored, the lack of a crucial "driver behavior" dimension results in a lack of correlation, semantics, and causal logic between the data. This prevents the formation of a complete closed-loop portrait of driving behavior—"driver perceives the environment → takes action → vehicle responds"—in other words, existing "vehicle" and "road" data are in a low-value and "dormant" state due to the absence of the "human" dimension, and the potential of automotive data has not been fully explored.

[0003] Existing vehicle data acquisition and processing technologies suffer from three major shortcomings. At the acquisition level, current systems primarily focus on vehicle status and environmental perception, either completely ignoring or recording key signals reflecting driver intentions and operating habits, such as brake pedal displacement and depress speed, accelerator pedal opening rate of change, and steering wheel angle and angular velocity, in isolation with extremely low sampling rates. This results in a "vehicle and road present, but no driver" data acquisition model. Furthermore, existing systems often employ continuous recording at fixed time intervals, generating a large amount of redundant data, while truly valuable transient driver behaviors are not prioritized for recording, resulting in extremely low data value density.

[0004] At the processing and analysis level, driver operation data, vehicle status data, and road environment data are collected and stored separately by different ECUs, lacking a unified timeline synchronization mechanism and failing to establish a fusion analysis framework with driver behavior data as the core dimension. The three are isolated from each other and stored separately, making it impossible to establish a complete causal chain of "environmental stimulus → driver operation → vehicle response".

[0005] At the storage and uploading level, the data generated by different sensors and ECUs have different formats and lack a unified data storage structure, resulting in low retrieval and reuse efficiency. At the same time, data uploading is mostly used for remote diagnosis or single-function feedback, lacking cloud-based collaborative design for data assetization, and failing to form a sustainable data asset pool. Summary of the Invention

[0006] To address the aforementioned deficiencies in existing technologies, this invention aims to systematically collect new driver behavior data and integrate it with road environment data and vehicle status data at the same time to form a "behavior-scenario-response" triple data unit. This establishes a complete causal chain of "environmental stimulus-driver operation-vehicle response," while simultaneously constructing an integrated architecture covering the entire chain of collection, synchronization, event-driven, fusion analysis, storage, and uploading. This significantly enhances the value of the original data by adding the "human" dimension.

[0007] To achieve the above objectives, the present invention adopts the following technical solution: A multi-source signal end-to-end processing system includes: a multi-source signal acquisition module 1, a multi-source data time synchronization and preprocessing module 2, an event-driven acquisition control module 3, a multi-source data fusion analysis module 4, a structured storage module 5, and a data upload and cloud collaboration module 6; the multi-source signal acquisition module 1 is connected to the multi-source data time synchronization and preprocessing module 2; the event-driven acquisition control module 3 is connected to the multi-source data time synchronization and preprocessing module 2; the multi-source data fusion analysis module 4 is connected to the event-driven acquisition control module 3; the structured storage module 5 is connected to the multi-source data fusion analysis module 4; and the data upload and cloud collaboration module 6 is connected to the structured storage module 5.

[0008] Furthermore, the multi-source signal acquisition module 1 includes a driver behavior acquisition unit, a road environment perception unit, and a vehicle status acquisition unit; the driver behavior acquisition unit outputs driver behavior data to the multi-source data time synchronization and preprocessing module 2, the road environment perception unit outputs road environment data to the multi-source data time synchronization and preprocessing module 2, and the vehicle status acquisition unit outputs vehicle status data to the multi-source data time synchronization and preprocessing module 2; the event-driven acquisition control module 3 receives driver behavior data from the driver behavior acquisition unit.

[0009] Furthermore, the driver behavior acquisition unit includes one or more of the following: brake pedal displacement sensor, brake pedal pressure sensor, accelerator pedal displacement sensor, accelerator pedal pressure sensor, and steering wheel angle sensor; the road environment perception unit includes one or more of the following: forward-looking camera, millimeter-wave radar, ultrasonic radar, and GPS module; and the vehicle status acquisition unit includes one or more of the following: wheel speed sensor, inertial measurement unit, and yaw rate sensor.

[0010] Furthermore, the event-driven acquisition and control module 3 includes an event determination rule base unit and an acquisition strategy control unit. The event determination rule base unit receives the driver behavior data from the multi-source signal acquisition module 1, and outputs an acceleration event signal when the accelerator pedal displacement change rate exceeds a first threshold, a braking event signal when the brake pedal displacement change rate exceeds a second threshold, and a steering event signal when the steering wheel angle or its change rate exceeds a third threshold. The acquisition strategy control unit receives the event signals, outputs a first acquisition frequency control signal to the multi-source signal acquisition module 1 when there are no event signals, and outputs a second acquisition frequency control signal higher than the first acquisition frequency and an extended acquisition time window control signal to the multi-source signal acquisition module when an event signal is received. The acquisition strategy control unit also outputs an event marker signal to the multi-source data time synchronization and preprocessing module 2.

[0011] Furthermore, the multi-source data fusion analysis module 4 includes a behavior-scenario-response pairing unit, a driving behavior label extraction unit, and a data value-added evaluation unit. The driving behavior label extraction unit receives a "behavior-scenario-response" triplet data packet, extracts one or more of braking style labels, following style labels, and economy labels based on the triplet data packet, and outputs the driving behavior labels to the structured storage module 5. The data value-added evaluation unit receives the "behavior-scenario-response" triplet data packet, generates one or more of data integrity indicators, dimension richness indicators, and available scenario quantity indicators, and outputs the value-added indicators to the structured storage module 5.

[0012] Furthermore, the structured storage module 5 includes a unified storage structure unit and an in-vehicle storage unit; the unified storage structure unit has a built-in raw data table, a synchronized data table, and an added-value data table; the raw data table receives and stores raw data from the multi-source data time synchronization and preprocessing module 2; the synchronized data table receives and stores multi-source data synchronized from the multi-source data time synchronization and preprocessing module 2, with timestamps as the primary key; the added-value data table receives and stores behavior-scenario-response triplet data packets from the multi-source data fusion analysis module 4 and their corresponding driving behavior tags and added-value indicators, with each triplet data packet having a unique identifier.

[0013] Furthermore, the data upload and cloud collaboration module 6 includes a data upload unit and a cloud collaboration unit. The data upload unit receives the "behavior-scenario-response" triple data packet and its corresponding driving behavior tags and value-added indicators from the structured storage module 5, and uploads it to the cloud collaboration unit through a 4G or 5G communication link, supporting breakpoint resume and encrypted data transmission. The cloud collaboration unit receives and aggregates triple data packets from multiple vehicles, performs cross-vehicle group driving behavior feature analysis, driving behavior prediction model training, and data asset registration, evaluation, and transaction management.

[0014] This invention also provides a multi-source signal end-to-end processing method, comprising the following steps: S1: Multi-source signal acquisition module 1 simultaneously acquires driver behavior data, road environment data, and vehicle status data; S2: Multi-source data time synchronization and preprocessing module 2 marks the collected data with a unified timestamp; S3: Event-driven acquisition and control module 3 sends an acquisition frequency switching command to multi-source signal acquisition module 1 based on the triggering events in the driver behavior data; S4: The multi-source data fusion analysis module 4 associates and fuses the driver behavior data, the road environment data, and the vehicle status data at the same time to generate a "behavior-scenario-response" triple data package; S5: The structured storage module 5 stores the "behavior-scenario-response" triple data packet into the value-added data table; S6: Data Upload and Cloud Collaboration Module 6 uploads the "behavior-scenario-response" triplet data package to the cloud server.

[0015] Furthermore, in step S1, the driver behavior data includes one or more of the following: brake pedal displacement and its rate of change, accelerator pedal displacement and its rate of change, and steering wheel angle; the road environment data includes the distance between the vehicle and the vehicle in front, relative speed, and road curvature radius; and the vehicle state data includes the vehicle speed, longitudinal acceleration, and yaw rate.

[0016] Compared with traditional solutions, the present invention has the following advantages: This invention adds driver behavior data as an incremental dimension to enhance data value. Unlike existing technologies that only collect environmental perception or vehicle status data, this invention integrates driver behavior data as a new dimension into the entire processing system, linking and fusing it with existing road environment and vehicle status data to form a complete causal chain of "behavior-scenario-response." This significantly increases the value of existing data by adding the "human" dimension, upgrading it from "what happened" to "who responded in what situation and how," thus activating dormant data.

[0017] This invention establishes a complete data link from acquisition, analysis, processing to storage and uploading, with clear interfaces between modules and smooth data flow. Compared to the fragmented approach of existing technologies, the end-to-end design of this invention ensures the integrity and consistency of data from generation to application, linking isolated data together.

[0018] The system of this invention transforms originally discrete, heterogeneous, multi-source raw data into value-added data assets with complete causal chains, high value density, and unique identifiers through an event-driven acquisition mechanism, a standardized "behavior-scenario-response" triplet storage structure, and cloud-based collaborative management. The system operates continuously during normal vehicle operation, constantly generating and accumulating such value-added data assets, forming a data resource pool that continuously appreciates in value over time, thus endowing static data with dynamic value.

[0019] The data collected by this invention simultaneously includes environmental stimuli, human behavior, and vehicle responses, forming a complete causal chain. This makes the data uniquely valuable in scenarios such as road improvement, traffic congestion management, and automotive product design enhancement.

[0020] This invention employs a trigger-based data collection mechanism centered on driver behavior events. It triggers high-frequency data collection only when high-value key behavioral events occur, avoiding redundant data storage in massive non-event states and significantly improving the data value density per unit of storage space.

[0021] Through the data upload and cloud collaboration module, the value-added data from the vehicle can be aggregated and analyzed across vehicles in the cloud to identify group behavior characteristics, train prediction models, and manage assets, thereby amplifying the value of data from individuals to groups.

[0022] This invention continuously collects and integrates driver behavior data with multi-source signals from "human-vehicle-road" sources, which can aggregate to form a massive database of real driving behavior covering different driver groups, different road types, different environmental conditions, and different traffic scenarios. Each data unit contains a complete causal chain of "environmental stimulus-driver behavior-vehicle response". This is the scarcest and most valuable data type in the training of autonomous driving algorithms, and it has reserved irreplaceable core data assets for the arrival of the autonomous driving era.

[0023] Based on the massive "behavior-scenario-response" triplet data generated by this invention, a large-scale driving behavior model with human-like driving ability, scenario generalization ability, and personalized style adjustment ability can be directly trained. At the same time, these real driving behavior data, especially stress response data, can serve as a "human reference standard" for the safety verification and iterative optimization of autonomous driving systems, and can be used for safety boundary definition, algorithm regression testing, and continuous iterative optimization, forming a data-driven positive evolutionary closed loop. Attached Figure Description

[0024] This manual includes the following figures, which illustrate the following: Figure 1 This is a block diagram of the overall hardware structure of the system of the present invention; Figure 2 This is an overall flowchart of the method of the present invention; The module includes: 1. Multi-source signal acquisition module; 2. Multi-source data time synchronization and preprocessing module; 3. Event-driven acquisition control module; 4. Multi-source data fusion and analysis module; 5. Structured storage module; and 6. Data upload and cloud collaboration module. Detailed Implementation

[0025] The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings, in order to help those skilled in the art to have a more complete, accurate and in-depth understanding of the concept and technical solutions of the present invention, and to facilitate its implementation.

[0026] like Figure 1 The diagram shown illustrates the system architecture of the multi-source signal end-to-end processing system of this invention, which integrates driver behavior data to drive data value-added processing. The system includes a multi-source signal acquisition module 1, a multi-source data time synchronization and preprocessing module 2, an event-driven acquisition control module 3, a multi-source data fusion and analysis module 4, a structured storage module 5, and a data upload and cloud collaboration module 6. The multi-source signal acquisition module 1 is connected to the multi-source data time synchronization and preprocessing module 2, transmitting its acquired driver behavior data, road environment data, and vehicle status data to the multi-source data time synchronization and preprocessing module 2. The event-driven acquisition control module 3 is connected to the multi-source data time synchronization and preprocessing module 2, receiving driver behavior data from the multi-source signal acquisition module 1, determining events, and outputting event marker signals to the multi-source data time synchronization and preprocessing module 2. The multi-source data fusion and analysis module 4 is connected to the event-driven acquisition and control module 3. It receives multi-source data synchronized from the multi-source data time synchronization and preprocessing module 2, as well as event marker signals from the event-driven acquisition and control module 3. After correlation and fusion, it generates a "behavior-scene-response" triplet data packet and outputs it to the structured storage module 5. The structured storage module 5 is connected to the multi-source data fusion and analysis module 4, and receives and stores the triplet data packets. The data upload and cloud collaboration module 6 is connected to the structured storage module 5, receives value-added data from the structured storage module 5, and uploads it to the cloud.

[0027] The multi-source signal acquisition module 1 includes a driver behavior acquisition unit, a road environment perception unit, and a vehicle status acquisition unit. The driver behavior acquisition unit includes a brake pedal displacement sensor, a brake pedal pressure sensor, an accelerator pedal displacement sensor, an accelerator pedal pressure sensor, and a steering wheel angle sensor. The brake pedal displacement sensor is installed at the brake pedal pivot and is used to acquire the brake pedal displacement signal S1; the brake pedal pressure sensor is installed at the brake pedal push rod and is used to acquire the brake pedal pressure signal N1; the accelerator pedal displacement sensor is installed at the accelerator pedal pivot and is used to acquire the accelerator pedal displacement signal S2; the accelerator pedal pressure sensor is installed at the accelerator pedal push rod and is used to acquire the accelerator pedal pressure signal N2; the steering wheel angle sensor is installed at the steering column and is used to acquire the steering wheel angle signal θ. Each sensor acquires data at a preset sampling frequency. The rate of change of brake pedal displacement dS1 / dt and the rate of change of accelerator pedal displacement dS2 / dt are obtained by differentiating their respective displacement signals with respect to time. The rate of change of steering wheel angle dθ / dt is obtained by differentiating the angle signal with respect to time. These differentiation operations can be performed in the signal processing circuit inside the sensor, or in the multi-source data time synchronization and preprocessing module 2. The road environment perception unit includes a forward-looking camera, millimeter-wave radar, ultrasonic radar, and a GPS module. The forward-looking camera collects lane line signals and traffic light status signals, and the millimeter-wave radar collects the distance signal D between the vehicle and the vehicle in front. rel and relative velocity signal V rel The GPS module collects vehicle location information and combines it with a high-precision map module to obtain the road curvature radius signal R. The vehicle status acquisition unit includes wheel speed sensors, an inertial measurement unit, and a yaw rate sensor. The wheel speed sensors collect wheel speed signals and calculate the vehicle speed signal V. host The inertial measurement unit acquires the longitudinal acceleration signal a x and lateral acceleration signal a y The yaw rate sensor collects the yaw rate signal ω. The driver behavior acquisition unit outputs driver behavior data, namely brake pedal displacement signal and its rate of change, brake pedal pressure signal, accelerator pedal displacement signal and its rate of change, accelerator pedal pressure signal, and steering wheel angle signal and its rate of change, to the multi-source data time synchronization and preprocessing module 2. The road environment perception unit outputs road environment data, namely the distance signal between the vehicle and the vehicle in front, relative speed signal, lane line signal, and road curvature radius signal, to the multi-source data time synchronization and preprocessing module 2. The vehicle status acquisition unit outputs vehicle status data, namely the vehicle speed signal, longitudinal acceleration signal, lateral acceleration signal, and yaw rate signal, to the multi-source data time synchronization and preprocessing module 2. The event-driven acquisition and control module 3 receives driver behavior data from the driver behavior acquisition unit for event determination.

[0028] The multi-source data time synchronization and preprocessing module 2 includes a time synchronization unit and a data preprocessing unit. The time synchronization unit receives a reference time signal from the GPS module or the vehicle clock, providing a unified timestamp reference for each acquisition unit in the multi-source signal acquisition module 1, ensuring accurate alignment of data across all channels in the time dimension, with time synchronization accuracy controlled within 10 milliseconds. The data preprocessing unit performs preprocessing operations on the acquired raw data: a low-pass filter is used to process the pedal signal to remove high-frequency noise; statistical methods are used to detect and remove outliers in each signal; and linear interpolation or spline interpolation methods are used to fill in missing data points and missing points in the original data after outlier removal.

[0029] The event-driven acquisition and control module 3 includes an event determination rule base unit and an acquisition strategy control unit. The event determination rule base unit receives driver behavior data, particularly displacement change rate data, from the driver behavior acquisition unit in the multi-source signal acquisition module 1 in real time, and performs event determination: when the accelerator pedal displacement change rate dS2 / dt exceeds a first threshold, it is determined as an acceleration event and an acceleration event signal is output; when the brake pedal displacement change rate dS1 / dt exceeds a second threshold, it is determined as a braking event and a braking event signal is output; when the steering wheel angle θ or its change rate dθ / dt exceeds a third threshold, it is determined as a steering event and a steering event signal is output. The first, second, and third thresholds can be calibrated according to different vehicle models and driving modes. For example, the first threshold can be set to 10% / s, the second threshold to 5% / s, and the angle threshold and change rate threshold in the third threshold can be set to 30° and 50° / s, respectively. If none of the above events are detected, the event determination rule base unit does not output an event signal. The data acquisition strategy control unit receives event signals from the event determination rule base unit and dynamically adjusts the acquisition strategy based on whether an event signal is received: when no event signal is received, it outputs a first acquisition frequency control signal to the multi-source signal acquisition module 1, controlling the multi-source signal acquisition module 1 to acquire data at the first acquisition frequency; when an event signal is received, it outputs a second acquisition frequency control signal to the multi-source signal acquisition module 1, controlling the multi-source signal acquisition module 1 to switch to a second acquisition frequency higher than the first acquisition frequency for data acquisition, and simultaneously outputs an extended acquisition time window control signal to ensure that the complete transient process before and after the event is recorded without losing key data; after the event signal disappears, the data acquisition strategy control unit restores the acquisition frequency from the second acquisition frequency to the first acquisition frequency after a preset delay. The data acquisition strategy control unit also outputs an event marker signal to the multi-source data time synchronization and preprocessing module 2. This marker signal includes the event type and the time of event occurrence, used to mark that the data within this time period contains driver behavior events.

[0030] The multi-source data fusion analysis module 4 includes a behavior-scenario-response pairing unit, a driving behavior label extraction unit, and a data value-added evaluation unit. The behavior-scenario-response pairing unit receives time-synchronized driver behavior data, road environment data, and vehicle status data from the multi-source data time synchronization and preprocessing module 2, as well as event marker signals from the event-driven acquisition and control module 3. Using timestamps as the primary key, it associates and pairs driver behavior data, road environment data, and vehicle status data at the same moment, generating a data structure. This data structure contains a behavior field (driver behavior data item at that moment), a scenario field (road environment data item at that moment), and a response field (vehicle status data item at that moment). These three elements combine to form a "behavior-scenario-response" triplet data packet. The pairing unit arranges all triplet data packets in chronological order and assigns a unique identifier to each triplet data packet before outputting it to the structured storage module 5. The driving behavior label extraction unit receives the triplet data packets and automatically extracts driving behavior labels based on the data contained within, including braking style labels, following style labels, and economy labels. Braking style tags categorize driver braking styles into smooth, standard, and aggressive types based on the statistical characteristics of brake pedal displacement change rate. Following style tags categorize following styles into safe, normal, and close types based on the statistical characteristics of distance and relative speed between the vehicle and the vehicle in front. Economy tags evaluate driving economy levels based on comprehensive indicators such as accelerator pedal opening change rate, braking frequency, and vehicle speed fluctuation. The data value-added evaluation unit receives triplet data packets and performs value-added evaluation on each triplet data packet: it calculates the missing rate of all fields in the triplet data packet to generate a data integrity index; it checks whether the triplet data packet contains complete dimensions of the three major data categories to generate a dimension richness index; and it automatically matches the number of application scenario types it can support based on the content of the triplet data packet to generate a usable scenario quantity index. After evaluation, the data value-added evaluation unit associates the value-added indicators with the corresponding triplet data packets and outputs them to the structured storage module 5.

[0031] The structured storage module 5 includes a unified storage structure unit and an on-board storage unit. The unified storage structure unit has three built-in data table structures and is managed using a relational database: a raw data table stores unprocessed raw sensor signals for data backtracking and secondary analysis; a synchronization data table stores time-synchronized multi-source fusion data, using timestamps as the primary key and sensor data as fields; and a value-added data table stores "behavior-scene-response" triplet data packets and their corresponding driving behavior tags and value-added indicators, with each triplet data packet having a unique identifier. The on-board storage unit uses solid-state drives as local storage media and supports time-based or capacity-based cyclic overwrite storage strategies. The structured storage module 5 also includes a data compression unit to perform lossless or lossy compression of stored data to save storage space.

[0032] The data upload and cloud collaboration module 6 includes a data upload unit and a cloud collaboration unit. The data upload unit uploads "behavior-scenario-response" triplet data packets, along with their corresponding driving behavior tags and value-added indicators, to the cloud server via a 4G or 5G communication module. The upload process supports breakpoint resumption and encrypted data transmission. The cloud collaboration unit, deployed on the cloud server, receives and aggregates triplet data packets from multiple vehicles, performing cross-vehicle group driving behavior feature aggregation analysis (identifying group driving behavior characteristics such as collective operation characteristics on specific road sections), driving behavior prediction model training (training driving behavior prediction models based on massive amounts of data), and data asset registration, evaluation, and transaction management. The data upload and cloud collaboration module 6 also includes a data interface unit, providing standardized API interfaces for secure and compliant data delivery to third-party data requesters.

[0033] like Figure 2 As shown, the present invention also provides a multi-source signal end-to-end processing method, comprising the following steps: S1: Multi-source signal acquisition module 1 simultaneously acquires driver behavior data, road environment data, and vehicle status data. Driver behavior data includes one or more of the following: brake pedal displacement and its rate of change, brake pedal pressure, accelerator pedal displacement and its rate of change, accelerator pedal pressure, and steering wheel angle and its rate of change. Road environment data includes one or more of the following: distance between the vehicle and the vehicle in front, relative speed, lane information, and road curvature radius. Vehicle status data includes one or more of the following: vehicle speed, longitudinal acceleration, lateral acceleration, and yaw rate.

[0034] S2: Multi-source data time synchronization and preprocessing module 2 marks the acquired data with a unified timestamp and performs filtering and noise reduction, outlier removal, and interpolation completion. The time synchronization accuracy is controlled within 10 milliseconds. Low-pass filters are used for filtering and noise reduction, statistical methods are used for outlier removal, and linear interpolation or spline interpolation is used for interpolation completion.

[0035] S3: The event-driven acquisition control module 3 sends an acquisition frequency switching command to the multi-source signal acquisition module 1 based on the trigger events in the driver behavior data. When no trigger event is detected, the multi-source signal acquisition module 1 operates at a first acquisition frequency; when a trigger event is detected, the multi-source signal acquisition module 1 operates at a second acquisition frequency higher than the first acquisition frequency, and after the event ends, it returns to the first acquisition frequency after a preset delay. Trigger events include acceleration events determined based on the accelerator pedal displacement change rate exceeding a first threshold, braking events determined based on the brake pedal displacement change rate exceeding a second threshold, and steering events determined based on the steering wheel angle or its change rate exceeding a third threshold.

[0036] S4: The multi-source data fusion analysis module 4 associates and merges driver behavior data, road environment data, and vehicle status data from the same moment to generate a "behavior-scenario-response" triplet data package. Specifically, using driver behavior data as the primary key, road environment data from the same moment is used as the scenario field, and vehicle status data is used as the response field for association and pairing, forming a triplet data package and assigning it a unique identifier. In the fusion analysis step, one or more of braking style labels, following style labels, and economy labels are extracted based on the triplet data package, and one or more of data integrity indicators, dimension richness indicators, and available scenario quantity indicators are generated.

[0037] S5: The structured storage module 5 stores the "behavior-scenario-response" triple data packets into the value-added data table, while storing the original data into the original data table and the synchronized data into the synchronized data table. Each triple data packet in the value-added data table has a unique identifier, forming a hierarchical storage management system and accumulative, reusable value-added data assets.

[0038] S6: Data Upload and Cloud Collaboration Module 6 uploads the "behavior-scenario-response" triplet data package and its corresponding driving behavior tags and value-added indicators to the cloud server. In the cloud, cross-vehicle group driving behavior feature aggregation analysis, driving behavior prediction model training, and data asset registration, evaluation, and transaction management are performed. The cloud server receives and aggregates triplet data packages from multiple vehicles, groups and statistically analyzes the data by road segment, identifies collective operational characteristics of specific road segments, and provides data support for road improvement and traffic management.

[0039] When a driver rapidly depresses the brake pedal due to a sudden deceleration of the vehicle in front, the event determination rule base unit detects in real time that the rate of change of brake pedal displacement exceeds the second threshold and immediately outputs a braking event signal. The acquisition strategy control unit then switches the acquisition frequency from low frequency to high frequency within 1ms, and extends the high-frequency acquisition by a preset time window before and after the event trigger moment. Through the above event-driven control process, the system effectively avoids the generation of a large amount of redundant data and the waste of storage space by acquiring data at low frequency when no event occurs, and automatically switches to high-frequency acquisition when a critical behavioral transient occurs to ensure that transient details are not lost, significantly improving the data value density per unit of storage space. At the same time, because the synchronization and preprocessing steps ensure the precise alignment of each data in the time dimension, the behavior-scenario-response pairing unit can accurately associate driver behavior data, road environment data, and vehicle status data at the same moment with timestamp as the primary key, generating a "behavior-scenario-response" triple data package containing a complete causal chain, providing complete semantic information that traditional data acquisition methods cannot provide for subsequent analysis.

[0040] After the system ran continuously on multiple test vehicles and accumulated a large number of triplet data packets, the cloud-based collaborative unit performed aggregation analysis, grouping and statistically analyzing the data by road segment. The large number of triplet data packets revealed abnormal distributions in indicators such as the frequency of driver sudden deceleration events, brake pedal change rate, and steering wheel angle change rate at specific road locations, with a significantly shortened time interval between accelerator and brake pedal switching. Based on this, the cloud-based collaborative unit generated congestion trigger point alarms and pushed them to traffic management departments. After on-site investigation, the locations of potential hazards such as unreasonable road design or obstructed visibility were identified, and the traffic conditions on these road segments were significantly improved after modifications. The triplet data packets output by the behavior-scenario-response pairing unit contain a complete causal chain of "environmental stimulus - driver operation - vehicle response," enabling the cloud to accurately pinpoint which road scenario characteristics led to which driver stressful operations and what abnormal vehicle responses, providing precise causal analysis evidence for road improvement that traditional data collection methods cannot offer.

[0041] This invention can also automatically extract driving behavior labels and generate value-added indicators for each triplet data packet. When training a driving behavior prediction model using the triplet data generated by this invention, the model can learn the true mapping relationship between scene features and driver behavior, rather than spurious correlations between data. The model trained with complete triplet data shows significantly better accuracy and generalization ability in predicting driver behavior than the model trained with only traditional "scene-response" binary data, especially when facing unfamiliar scenes not present in the training set. Through the integrated processing of the entire chain of acquisition, synchronization, event-driven, fusion analysis, storage, and uploading, this invention significantly enhances the value of raw data by adding the "human" dimension, and has broad industrial application value.

[0042] The present invention has been described above by way of example with reference to the accompanying drawings. Obviously, the specific implementation of the present invention is not limited to the above-described manner. Any non-substantial improvements made using the inventive concept and technical solution; or the direct application of the inventive concept and technical solution to other situations without modification, are all within the protection scope of the present invention.

Claims

1. A multi-source signal end-to-end processing system, characterized in that, include: The system comprises a multi-source signal acquisition module (1), a multi-source data time synchronization and preprocessing module (2), an event-driven acquisition control module (3), a multi-source data fusion analysis module (4), a structured storage module (5), and a data upload and cloud collaboration module (6). The multi-source signal acquisition module (1) is connected to the multi-source data time synchronization and preprocessing module (2). The event-driven acquisition control module (3) is connected to the multi-source data time synchronization and preprocessing module (2). The multi-source data fusion analysis module (4) is connected to the event-driven acquisition control module (3). The structured storage module (5) is connected to the multi-source data fusion analysis module (4). The data upload and cloud collaboration module (6) is connected to the structured storage module (5).

2. The multi-source signal end-to-end processing system according to claim 1, characterized in that, The multi-source signal acquisition module (1) includes a driver behavior acquisition unit, a road environment perception unit, and a vehicle status acquisition unit; the driver behavior acquisition unit outputs driver behavior data to the multi-source data time synchronization and preprocessing module (2), the road environment perception unit outputs road environment data to the multi-source data time synchronization and preprocessing module (2), and the vehicle status acquisition unit outputs vehicle status data to the multi-source data time synchronization and preprocessing module (2); the event-driven acquisition control module (3) receives the driver behavior data from the driver behavior acquisition unit.

3. The multi-source signal end-to-end processing system according to claim 2, characterized in that, The driver behavior acquisition unit includes one or more of the following: brake pedal displacement sensor, brake pedal pressure sensor, accelerator pedal displacement sensor, accelerator pedal pressure sensor, and steering wheel angle sensor; the road environment perception unit includes one or more of the following: forward-looking camera, millimeter-wave radar, ultrasonic radar, and GPS module; and the vehicle status acquisition unit includes one or more of the following: wheel speed sensor, inertial measurement unit, and yaw rate sensor.

4. The multi-source signal end-to-end processing system according to claim 1, characterized in that, The event-driven acquisition control module (3) includes an event determination rule base unit and an acquisition strategy control unit. The event determination rule base unit receives the driver behavior data from the multi-source signal acquisition module (1), and outputs an acceleration event signal when the accelerator pedal displacement change rate exceeds a first threshold, an acceleration event signal when the brake pedal displacement change rate exceeds a second threshold, and a steering event signal when the steering wheel angle or its change rate exceeds a third threshold. The acquisition strategy control unit receives the event signals, outputs a first acquisition frequency control signal to the multi-source signal acquisition module (1) when there are no event signals, and outputs a second acquisition frequency control signal higher than the first acquisition frequency and an extended acquisition time window control signal to the multi-source signal acquisition module (1) when the event signals are received. The acquisition strategy control unit also outputs an event marker signal to the multi-source data time synchronization and preprocessing module (2).

5. The multi-source signal end-to-end processing system according to claim 1, characterized in that, The multi-source data fusion analysis module (4) includes a behavior-scenario-response pairing unit, a driving behavior label extraction unit, and a data value-added evaluation unit. The driving behavior label extraction unit receives the "behavior-scenario-response" triplet data packet, extracts one or more of the braking style label, following style label, and economy label based on the triplet data packet, and outputs the driving behavior label to the structured storage module (5). The data value-added evaluation unit receives the "behavior-scenario-response" triplet data packet, generates one or more of the data integrity index, dimension richness index, and available scenario quantity index, and outputs the value-added index to the structured storage module (5).

6. The multi-source signal end-to-end processing system according to claim 1, characterized in that, The structured storage module (5) includes a unified storage structure unit and an in-vehicle storage unit; the unified storage structure unit has a built-in original data table, a synchronization data table and an added value data table; the original data table receives and stores the original data from the multi-source data time synchronization and preprocessing module (2); the synchronization data table receives and stores the time-synchronized multi-source data from the multi-source data time synchronization and preprocessing module (2), with the timestamp as the primary key; the added value data table receives and stores the behavior-scenario-response triplet data packets and their corresponding driving behavior tags and added value indicators from the multi-source data fusion analysis module (4), and each triplet data packet has a unique identifier.

7. The multi-source signal end-to-end processing system according to claim 1, characterized in that, The data upload and cloud collaboration module (6) includes a data upload unit and a cloud collaboration unit. The data upload unit receives the "behavior-scenario-response" triple data packet and its corresponding driving behavior tags and value-added indicators from the structured storage module (5), and uploads it to the cloud collaboration unit through a 4G or 5G communication link, and supports breakpoint resume and encrypted data transmission. The cloud collaboration unit receives and aggregates triple data packets from multiple vehicles, performs cross-vehicle group driving behavior feature analysis, driving behavior prediction model training, and data asset registration, evaluation and transaction management.

8. A processing method for a multi-source signal end-to-end processing system according to any one of claims 1-7, characterized in that, Includes the following steps: S1: The multi-source signal acquisition module (1) simultaneously acquires driver behavior data, road environment data and vehicle status data; S2: The multi-source data time synchronization and preprocessing module (2) marks the collected data with a unified timestamp; S3: The event-driven acquisition control module (3) sends an acquisition frequency switching command to the multi-source signal acquisition module (1) based on the triggering event in the driver behavior data; S4: The multi-source data fusion analysis module (4) associates and fuses the driver behavior data, the road environment data and the vehicle status data at the same time to generate a "behavior-scenario-response" triple data package; S5: The structured storage module (5) stores the "behavior-scenario-response" triple data packet into the value-added data table; S6: The data upload and cloud collaboration module (6) uploads the "behavior-scenario-response" triple data packet to the cloud server.

9. A multi-source signal end-to-end processing method according to claim 8, characterized in that, In step S1, the driver behavior data includes one or more of the following: brake pedal displacement and its rate of change, accelerator pedal displacement and its rate of change, and steering wheel angle; the road environment data includes the distance between the vehicle and the vehicle in front, relative speed, and road curvature radius; and the vehicle state data includes the vehicle speed, longitudinal acceleration, and yaw rate.

Images