Energy Storage Systems
Technical Dialogue Session
Renewable Energy Systems II
lithium-ion battery, life cycle, state-of-health, deep learning, regression model, battery management system, real-time detection
This paper presents a data-driven regression model to predict the life cycle of lithium-ion battery. The model is built based on five different key features derived from discharging voltage, current, time, and skin temperature. These sets of data are commonly available in most Battery Management Systems (BMS), which makes the proposed regression model widely applicable to various battery systems. The five-key-feature-based battery model is evaluated and validated all through the paper and is proved to have high prediction accuracy of battery life cycles. The proposed regression model can be utilized for real-time life cycle prediction without the need of historical data.