| With the increased use of Lithium ion batteries in a variety of applications, the presence of an anomaly proves to be a major concern as it not only affects the battery, but also affects the battery operated system. Battery Management System (BMS) can be equipped with various anomaly detection procedures to detect failures and attacks and hence prevent improper functioning and catastrophic events caused by such anomalies. In this research, the Lithium ion battery is modeled into a first order RC equivalent circuit to understand its behavior. Kalman filter is used to estimate the states and an adaptive estimation algorithm is used to estimate the model parameters. Residual based detection mechanism is employed for anomaly detection. By understanding the performance of the detectors and comparing them with each other, they are tuned to detect the zero-alarm attacks which equip them for worst-case attack detection. |
