Neural Architecture Search-Guided Physics-Informed Neural Network for Energy Management in Hybrid Energy Storage System with Electric Vehicles
M. Sivaramkrishnan, L Ancelin, M. Siva Ramkumar, Osama Jamal Jamil Abed Al Jawad, Jayant Giri, Tunga Vignesh
Abstract
An efficient Energy Management (EM) of a Hybrid Energy Storage System (HESS) combining batteries and Supercapacitors (SCs) is essential for enhancing the performance and reliability of Electric Vehicles (EVs). However, challenges such as high integration costs, energy losses during power conversion, and the complexity of coordinating multiple storage units can impact overall system efficiency and longevity. To overcome these drawbacks, this manuscript proposes a technique for optimizing EM in HESS with EVs. The suggested method is named as Neural Architecture Search-Guided Physics Informed Neural Network (NASPINN). The main aim of the suggested EM system is to enhance the integration of a HESS for EVs, improving efficiency, reliability, and sustainability while reducing operational costs and maximizing energy utilization. The suggested NASPINN predicts future energy demand and charging/discharging profiles for EVs based on driving patterns and power input variations. By then, the suggested approach is implemented in MATLAB and contrasted with several other methods that are currently in use. The suggested technique outperforms all existing approaches such as Fuzzy Logic Control (FLC), Artificial Neural Network (ANN), and Convolutional Neural Network (CNN). The GAO-MSNN technique reaches operational efficiency at 98.7% while using an operational cost that equals 1503cents. These results show that the suggested method leads existing techniques in both operational efficiency and reduced cost while presenting itself as an optimal option for HESS optimization in EVs.