Long-Short Term Memory Model to Estimate Remaining Operational Time of Battery Cycles in Electric Vehicles Through Hyper Parameter Analysis

Among available technologies in the growing adoption of electric vehicles (EVs), lithium-ion batteries (LiBs) have emerged as the preferred solution due to their superior energy density, high voltage output, and low self-discharge characteristics. This work investigates the application of Long Short-Term Memory (LSTM) neural networks for predicting the Remaining Useful Life (RUL) of LiBs. A key focus of this work is the analysis of voltage decrement during discharge cycles, specifically 3.6 V to 3.4V transition period, which serves as an important degradation indicator. The LSTM model demonstrates exceptional capability in identifying complex nonlinear degradation patterns from this voltage feature, achieving high-precision RUL predictions that show strong alignment with experimental data. The findings reveal that (1) Increasing hidden node count enhances the model's capacity to learn intricate battery degradation patterns and (2) Longer input sequence lengths improve temporal feature extraction, capturing essential long-term dependencies. These insights provide valuable guidance for developing robust battery management systems, contributing to improved EV performance and longevity.