Neural network based SOC estimation during cycle aging using voltage and internal resistance

Abstract

The state of charge(SOC) estimation of a Lithium-ion battery is critical and fundamental in managing and using it safely and optimally. The SOC estimation often overlooks the effects of aging and cycling of the battery. This paper presents a model free estimation method based on the Neural network algorithm using the terminal voltage and the internal resistance instead of the capacity. The terminal voltage plays a role in the SOC estimation of each cycle and the internal resistance adapts it according to the cycle aging effect. After the training of the network for a few intermittent cycle numbers, the SOC estimation error with different cycle numbers for testing results in less than 3 percent. This method can be used in the EV battery applications in which the model parameters are not easy to obtain.