Quantum Annealing-Based Machine Learning for Battery Health Monitoring Robust to Adversarial Attacks

Battery health monitoring methods including machine learning (ML) models rely on trustworthiness of battery sensor data and features. As more battery systems require network connectivity for intelligent health monitoring, cyberattacks become a serious threat. A notable attack is an adversarial attack that generates stealthier input data misleading the ML results. This paper introduces an adversarial attack targeting battery health monitoring system and explores a quantum annealing (QA)-based machine learning method to mitigate the adversarial attacks. In this paper, a Quantum Boost (QBoost)-based state of health (SOH) estimation algorithm for electric vehicles is developed and trained by D-Wave QA processor and the performance is compared with a state-of the art random forest (RF)-based SOH estimation method. The experimental results show that the proposed QBoost method is more robust to the adversarial attack than the RF method, while the RF method provides better accuracy during normal operation.