Time and Frequency Domain Health Indicators for Capacity Prediction of Lithium-ion Battery

Abstract

Predict the capacity of lithium-ion batteries with high accuracy is crucial to the reliability and safety of the system. Due to the complex nature and the nonlinear degradation phenomena of the lithium-ion battery, monitoring the battery's capacity is a challenging task. This paper proposes a machine learning model based on time and frequency domain health indicators to predict the capacity of lithium-ion battery cycled under different operational conditions. The time and frequency domain health indicators have been extracted from the measured voltage. The extracted features have been fed into extreme learning machine model to predict the capacity. This approach has been tested on 16 lithium-ion batteries cycled at many operational conditions from NASA. The results show that the proposed method can track the degradation from the extracted health indicators from both domains (time and frequency). The extreme learning model can effectively predict the capacity with a root mean square error of 1.3%.