Battery state of charge non-smooth hysteresis characteristic compensation estimation method and system

The present invention is a battery-powered non-smooth hysteresis characteristic compensation charge state estimation method and system, the first step in the acquisition of this Law cell output voltage and current obtained from a battery equivalent circuit model parameters of the relationship to build neural networks OCV (k) pre assessment model, which solved parameters, open circuit voltage OCV (k) estimate online. Step SDH model and the dynamic hysteresis RBF2 in series hybrid model. Step SDH model resulting OCV (k) is the input, y (k) and the OCV (k) of the output, OCV (k-1) is input RBF2, RBF2 indirect learning weighting parameter adjustment SDH model, to approach real SOC (k) of the complex relationship between the delay, the final output line estimation. The system consists of a microprocessor and attached to the battery circuit current, voltage sensors, and stores a program of the present method of execution to give SOC (k) estimates. Neural networks learn from the present invention compensates for battery complex non-smooth hysteresis characteristics, improve SOC (k) line estimation accuracy.