Resilient Filter for State of Charge and Parameter Coestimation With Missing Measurement

Accurate state of charge (SOC) can effectively improve safety performance and prolong the cycle life of the batteries. The widely used model-based SOC estimation methods have underlying assumptions of complete measurements and accurate estimator gains, which are not always reasonable in practical applications. Thus, this article designs a dual Kalman filter-type resilient filter to estimate SOC and parameter jointly with the random missing measurement phenomenon which is modeled by a Bernoulli distributed sequence. Besides, the filter gain variations, in both online parameter identification and state estimation, are characterized by mutually independent multiplicative noise terms. Then, based on the minimum-variance principle, the filter gains are designed to minimize the effects of the missing measurement and gain variations on the estimation performance. Finally, extensive simulations and experiments are conducted to validate the effectiveness and resilience of the proposed method.