As lithium-ion batteries are increasingly used as the main power source for large-capacity applications, the life prediction system based on the Battery Management System (BMS) is becoming more important. Considering the various environments of applications in which lithium-ion batteries are mounted and operated, a universal battery management technology is also required. In order to develop battery management technology, it is essential to select the indicators according to various conditions, and it must be applicable according to variable circumstances. Therefore, an electrical characteristic experiment was conducted by selecting a profile that simulates the driving environment of an electric vehicle, and an aging trend was analyzed. In order to improve the performance of the artificial intelligence-based State-of-Health (SOH) prediction model, proposed labeling technique of indicators according to various conditions can be used. Labeled multi-health indicators are used as input data for prediction model. The prediction model estimates the SOH suitable for variable situations based on data pre-processing, and verifies the model performance using different cell data.
