Remaining Useful Life Prediction of Super-Capacitors in Electric Vehicles Using Neural Networks

Abstract

Batteries for electric vehicles (EVs) have a capacity decay issue as they age. As a result, the use of lithium-ion is becoming more popular with super-capacitors (SCs), particularly in EVs. Over the decrease of carbon dioxide emissions, SC batteries offer a substantial benefit. In EVs, a dependable mechanism that guarantees the SC batteries’ capacity for charging and discharging is crucial. The main obstacle for EVs is the long life of ultra-capacitor battery’s because SCs have a deterioration effect over multiple cycles. Therefore, accurate early prediction of these SC batteries is crucial. The data-based model is more accurate than mechanism-based and model-based methods created for this purpose. The proposed data-driven models, such as machine learning (ML), estimate the electrical parameters for the smooth functioning and working of SCs in addition to considering their operating status. The main factor determining whether electric vehicles can be sustained is an increase in battery cycle life. With a lowest root mean square error of 0.04614 and a mean squared error of 0.002 and an accuracy of 89.6%, ML-based models with various architectures and topologies have been created in this study to reliably estimate the deterioration of SCs capacitance.