Next-Gen Electric Vehicle Battery Optimization: Enhancing RUL Prediction with Regression-Based Algorithms

Batteries have become the central component of energy storage frameworks, playing a crucial role in supplying power to various devices and technologies. Among this, rechargeable electrochemical storage or batteries made with lithium-ion have a high density of energy while offering a lightweight structure and relatively low self-discharge rates. They have emerged as an essential component of modern battery backup systems, especially in electric vehicles (EVs). However, concerns related over their limited lifespan and continuous, subtle decline within the battery capacity, and long-term reliability pose barriers to their widespread adoption. This research work focuses on enhancing the estimation of RUL (Remaining Useful Life) through the utilization of data-driven techniques. These techniques are growing increasingly and are essential for battery management in vital industrial sectors such as electric vehicles and solar energy storage. Although traditional machine learning models are effective in certain circumstances, they frequently lack the robustness and reliability required for accurate RUL prediction. To address these challenges, we are utilizing Random Forest(RF), LightGBM (LGBM), and a hybrid of these models to boost prediction accuracy and reliability. The hybrid model performs more effectively in reducing errors and improving predictive accuracy, eliminating the limitations of individual algorithms, which was tested on datasets that involve lithium-ion batteries. In accordance with our research, this approach provides a more reliable way of estimating the health of batteries, contributing to safer battery management systems whilst being effective.