A dual-rate sampled multiple innovation adaptive extended Kalman filter algorithm for state of charge estimation

For a lithium battery state space model, to solve the unsuitable problems caused by the single-rate (same rate) sampled state space model facing resistor-capacity (R-C) couples with different time scales, a dual-rate sampled exponential weighted multiple innovation adaptive extended Kalman filter (AEKF) method is investigated. The details include: (a) By using dual-rate sampled method, select the suitable sampled rate for their own time scale for different couples (R1-C1 and R2-C2) to establish dual-rate sampled state space equations; (b) Considering noise variation, a recursive noise estimation based AEKF algorithm is adopted to realize adaptive correction of noise; (c) Considering the influences of present and past data, an AEKF algorithm based exponential weighted multiple innovation is studied. Experiment results show that the investigated method is clearly better than the classical single-rate sampled method with high precision.