Enhancing Kalman Filter Resilience in Electric Vehicles: Cyber-Attack Mitigation with Machine Learning Based Adaptive Filtering

Electric vehicles (EVs) depend on Kalman Filters for precision in navigation and control; however, these systems can be exposed to cyber-attacks such as Denial-of-Service (DoS) and data injections, which can compromise sensor data integrity. This study presents a unique approach that combines Adaptive Kalman Filters with Isolation Forest algorithm to detect real-time anomaly and adjust the parameters to improve system resilience. Simulation-based experiments were conducted, demonstrating that the Adaptive Kalman Filter can reduce the Root Mean Square Error (RMSE) by 40% and Mean Absolute Error (MAE) by 35% during cyber-attacks compared to Standard Kalman Filters. The findings from this study illustrate that implementing both Adaptive Kalman Filters with machine learning-based anomaly detection methods can significantly improve the cyber resilience of electric vehicles and strengthen the development of a stronger state estimation framework, improving the cybersecurity and reliability of EV sensor fusion systems.