A Synergistic Approach using FCN and Gradient Descent Optimization for RUL Prediction of Lithium Ion Batteries

This research compares the performance of a gradient descent-optimized algorithm with a hybrid technique that combines gradient descent and Fully Connected Networks (FCN) to determine the effectiveness of this method for forecasting the Remaining Useful Life (RUL) of lithium-ion batteries. Utilizing a dataset of around 7,000 records, the analysis demonstrates that the gradient descent-optimized linear regression model exhibits strong predictive capabilities, achieving a Mean Squared Error (MSE) of 0.2437 and an R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> of 0.9497, reflecting a robust fit to the data. However, the combined model, which integrates gradient descent with FCN, significantly outperforms the standalone model. It reduces the MSE to 0.0823 and enhances the R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> to 0.9830, showcasing superior accuracy in predicting battery lifespan. These findings underscore the potential of hybrid models to elevate predictive precision for RUL forecasting in lithium-ion batteries, offering valuable insights into model optimization strategies for effective battery health monitoring.