High-Risk Electric Vehicle Identification Based on Logistic Regression

As the primary energy storage component in electric vehicles, the reliability of lithium-ion batteries is of paramount importance. Identifying high-risk vehicles is crucial to ensure the safety of electric vehicles and their users. Traditional fault diagnosis methods predominantly depend on the real-time collection of battery status parameters by the onboard Battery Management System (BMS) to facilitate diagnostics and trigger alert notifications. However, these approaches suffer from inherent latency issues and have limited ability in predicting potential risks. Furthermore, existing methods for extracting the features of risks and utilizing big data techniques for fault diagnosis have not established precise classification boundaries. To overcome these limitations, this paper introduces an innovative fault diagnosis approach, which entails modeling of various abnormal battery behaviors, followed by the creation of precise mathematical expressions to quantitatively represent each of these risk behaviors. Subsequently, leveraging actual operational data from electric vehicles collected by the National Monitoring and Management Center for New Energy Vehicle (NMMCNEV), this study employs advanced machine learning algorithms, such as Logistic Regression algorithm, to calculate customized risk factors for real vehicles and optimize the parameters of a multi-feature input model. Validation results confirm the feasibility and robustness of the proposed fault diagnosis method, indicating its capability to complement traditional fault diagnosis approaches.