Abstract:

The accurate estimation of state of charge (SOC) of lithium-ion cells is the key technology of battery management system. The influence of SOC estimation accuracy with the drift or white noise of voltage or current sensor was analyzed to determine the sensor errors on the accuracy of SOC estimation using genetic algorithm to identify model parameters and extended Kalman filter based on the second-order RC equivalent circuit model. The results show that the relationship between offset values and the mean of SOC estimation error has an approximately linear when the voltage or current sensor has drift, but the white noise of the voltage and current sensors has no influence on the mean of the SOC estimation error. If the estimation accuracy of SOC is less than 5% for lithium-ion cell used in the experiment, the voltage and current offset should be controlled respectively within 10 mV and 1/30 C.

Key words: ternary lithium-ion cells, sensor error, state of charge (SOC), extended Kalman filter, battery equivalent circuit model, parameter identification