Novel data-driven model for estimating lithium-ion battery health in electric vehicles using sensor data

An efficient evaluation of lithium-ion battery (LIB) health is crucial for certifying the consistency, safety, and durability of electric vehicles (EVs). Traditional methods often rely on intrusive testing or simplified models, which fail to imprison the complex, nonlinear behavior of batteries in dynamic environments. To address these limitations, a new data-driven approach using Adaptive Artificial Fish Swarm Algorithm-driven Restricted Boltzmann Machines (AAFSA-RBM) is proposed. This advancement presents an effective method for estimating battery health by leveraging sensor data, including temperature, voltage, current, and state-of-charge (SOC). Pre-processing methods like min-max normalization are used to scale the data, ensuring all values fall within a specified range. Principal Component Analysis (PCA) is employed to reduce redundancy and preserve the most significant features of the data, improving computational efficiency. The AAFSA-RBM method optimizes the model parameters, significantly improving the accuracy of battery health predictions. Significant gains over conventional techniques are demonstrated by the results of the suggested model, which show mean square error (MSE) of 0.021, root mean square error (RMSE) of 0.010, and mean absolute error (MAE) of 0.009. This contemporary advance provides a breakthrough solution for estimating EV battery health. It offers a non-invasive, scalable, and accurate tool for monitoring the health of LIBs in real time, helping to improve maintenance strategies, develop battery longevity, and make certain the safe function of electric vehicles. This method attends to the critical needs of modern EV systems. It paves the way for more reliable and efficient battery management, facilitating the broader adoption of electric vehicles while improving their overall sustainability.