Battery state-of-charge estimation based on chaos immune evolutionary neural network

It is difficult for conventional method to correctly predict battery state of charge (SOC). Partial least square regression is used to select input variables. Battery terminal voltage, voltage derivative, voltage second derivative, discharge current and battery temperature are inputs of neural network and state of charge is output. In order to avoid defect of BP algorithm, the adaptive parallel chaos immune evolutionary programming was adopted to train weights. Finally, under the state of dynamic power cycle, predicted SOC and actual SOC were compared to verify the proposed neural network within 5 percent error.