Pre-Processing Measurement Data for Computing Internal DC Resistance with Anomaly Detection Techniques

A significant concern in using electric vehicles (EVs) is the range variability, despite vehicles being from the same manufacturers and driven under similar conditions. This vari-ance often stems from cell-to-cell impedance variation within the battery pack. Since the impedance is a very small value ranging in milliohm (mO), the measurement requires precise signal generation and measurement, followed by data post-processing to ensure accurate outcomes. The focus of this paper is to propose the use of anomaly detection techniques in preprocessing the raw data to ensure reliable results are obtained when using the data. In this study, three anomaly detection methods were examined and their precision and limitations in preprocessing measured data were assessed. Based on the evaluation and conclusions drawn, the best method among these three was chosen for anomaly detection during preprocessing. Analysis of Variance (ANOVA) is employed on the healthy data of the internal DC resistances to explore patterns in the variation with respect to manufacturers.