Noise diagnosis of commercial Li-ion batteries using high-order moments

Abstract

Electrochemical noise diagnostics (END) has been applied to commercial Li-ion batteries. This diagnostics is based on recording small voltage fluctuations of the battery during discharge. In order to avoid apparatus noise the direct measurements of voltage have been used without filtering and statistical descriptors of process have been calculated, namely standard deviation (STD), skewness (SK) and kurtosis (KU). It was demonstrated, as a function of State of Charge (SOC), that the noise increases drastically at the beginning (SOC > 97%) and the end (SOC < 4%) of the discharge. The noise is quasi-uniform and non-Gaussian with SK = 0 and KU =–1.2. In the boundaries between the beginning and the middle and between the middle and the end, an important variations of SK and KU have been detected.