Predicting battery degradation profiles of IoT device usage modes through Machine Learning utilization models

IoT-enabled devices powered by rechargeable batteries often need to be compact, which results in using low-capacity batteries that require frequent charging. The high cycle count leads to noticeable battery health degradation. Therefore, understanding how different usage modes of IoT devices affect the State of Health (SoH) of the battery is crucial. Energy consumption and battery degradation vary significantly across usage modes, making it important to identify the specific impact each mode has on the battery health. Traditional lab-based testing is often time-consuming, costly, and may not accurately reflect real-world usage patterns. This paper introduces a systematic procedure that leverages established Machine Learning (ML) utilization models to reveal degradation profiles associated with various usage modes of IoT devices and their impact on battery health. These degradation profiles illustrate how the runtime in a given usage mode influences the SoH of a battery. This insight allows product developers to optimize existing usage modes and improve the efficiency of next-generation IoT products. Additionally, these degradation profiles enable accurate predictions of the battery’s Remaining useful Lifetime (RuL) by providing a measure for the decrease in SoH based on the runtime of a specific usage mode. • ML approach to predict battery degradation profiles of IoT device usage modes. • Degradation profiles illustrate runtime impact on battery State of Health (SoH). • Normalized degradation profiles ensure comparability across modes/batteries. • ML model trained on IoT data predicts battery degradation accurately. • ML-based profiles outperform linear regression in predicting degradation.