Machine Learning Based Battery Anomaly Detection Using Empirical Data

Abstract

In the context of energy storage systems (ESS), this work investigates the use of machine learning approaches for anomaly identification utilizing empirical site data. Making advantage of the empirical data gathered from the operational environment, the study concentrates on using precise anomaly detection techniques-mainly the Isolation Forest method. The Isolation forest approach is utilized to detect abnormalities in the empirical data obtained by ESS operations. It is well-known for its effectiveness in locating outliers in datasets. In order to improve the operational dependability and safety of Energy Storage Systems (ESS), this study explores the application of the Isolation Forest technique as a powerful tool for identifying anomalies in the site data. The results of the study show that, Isolation forest can detect anomalies with the accuracy of 99.43 %.