一种多厂商车辆数据通用预处理方法、装置及电子设备
By using ID whitelists and DBC configuration file parsing rules in vehicle data processing, automated preprocessing of data from different manufacturers is achieved, solving the inefficiency problem caused by inconsistent data formats and supporting rapid migration and efficient training of cross-vehicle models.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- LIUZHOU WULING NEW ENERGY VEHICLE CO LTD
- Filing Date
- 2026-05-13
- Publication Date
- 2026-07-17
AI Technical Summary
In vehicle R&D and testing, the inconsistent CAN bus data formats of different manufacturers lead to low efficiency, inconsistent quality and poor universality in data model construction. Technicians spend a lot of time on tedious data preprocessing, which hinders the rapid iteration of models and cross-vehicle generalization.
By reading the raw log files, filtering messages using an ID whitelist, and combining the parsing rules of the DBC configuration file, signal decoupling and resampling are achieved, constructing structured vehicle data, and automating the transformation from raw data to model input.
Significantly reduces data processing volume, decouples signal definition and processing logic, supports cross-platform and cross-vehicle model migration, provides high-quality, standardized training data, and improves model development efficiency.
Smart Images
Figure CN122412401A_ABST