Integration of Hybrid Energy Storage System and Photovoltaic with Improving Microgrid Performance using Supervised Attention Multi-Scale Temporal Convolutional Network

Integrating Battery Energy Storage Systems (BESSs), Supercapacitor (SCs) and Photovoltaic (PV) panels into Microgrid (MGs) increases reliability through energy regulation and stabilization. However, the acquisition cost of the device is rather high, and it has constant maintenance costs, which can lead to a higher total annual cost. Furthermore, the deterioration of these storage components over time affects both the ability of the systems and the overall cost. In an attempt to address these drawbacks, this manuscript presents an optimal operating strategy of MG system with HESS and PV. The proposed method is Supervised Attention Multi-Scale Temporal Convolutional Network or SAMTCN for short. The general objective of the proposed method is to minimize the TAC of the MG system and to enhance the overall performance of the system. The SAMTCN that is proposed in this paper is employed to forecast the energy demand and supply trends prevailing in the MG, in order to fully exploit the HESS and PV panels used. Through the precise projection of energy variations, SAMTCN minimizes wastage of energy, energy storage losses are avoided, and general system efficiency is enhanced. Then, the proposed technique is with implemented in MATLAB and compared with other techniques that have been put forward. The proposed method outperforms all existing approaches such as Artificial Neural Network (ANN), Improved Deep Neural Network (IDNN), Fuzzy Logic Control (FLC), Artificial Rabbits Optimized Neural Network (ARONN), and Multi-objective Artificial Hummingbird Algorithm (MOAHA). The results demonstrate that SAMTCN outperforms these methods, achieving a TAC of $132,270.8 and an efficiency of 99.7%. These outcomes confirm the superiority of SAMTCN in reducing costs and maximizing the efficiency of MG systems with HESS and PV panels.