Performance Diagnostics in Photovoltaic-Lithium-Battery Installations using Inconsistent Field Data

Performance diagnostics of batteries in solar-photovoltaic and battery systems are important, especially if using second-life electric vehicle batteries. Currently, the battery pack is often oversized and suboptimally managed. This is partially due to the lack of high-quality data collection and processing on-site. Practical implementation of advanced diagnostics tools such as machine learning remains out
of reach. Here a robust pulse-based diagnostics method is proposed for calculating the equivalent series resistance of the battery pack. It takes advantage of ‘natural’ pulse instances in the system. Feature selection of the voltage and current measurements is combined with simple regression techniques to obtain an estimate. Data from 5 real-world installations, split into 9 distinct datasets, is used to assess the method. While there was no opportunity to compare the estimates with the true resistance values, the consistency of the results suggests that our method may be generally applicable.