ENHANCING BATTERY THERMAL MANAGEMENT IN ELECTRIC VEHICLES：A HYBRID DMCOA ALGORITHM AND DEEP NEURAL NETWORK APPROACH

Abstract

This paper presents an innovative approach to Battery Thermal Management Systems (BTMS) utilizing a hybrid algorithm, the Dwarf Mongoose-based Coati Optimization Algorithm (DMCOA), in conjunction with a deep neural network (DNN). Our objective is to optimize the temperature of lithium-ion batteries, particularly in Electric Vehicles (EVs). The DMCOA draws inspiration from cooperative behaviors seen in coatis and dwarf mongooses. It employs advanced strategies, such as cooperative attacks simulation and escape behavior imitation to ensure efficient minimization of cost function. Additionally, a DNN is employed to predict vehicle speed and battery heat production rate under various conditions, enhancing the control of the BTMS. Simulation outcomes demonstrate the effectiveness of the hybrid algorithm in maintaining battery temperatures, with minimal deviation from the target range. Simulation results show that the proposed hybrid algorithm efficiently maintains battery temperatures, with just a 0.3°C average difference from the target and a maximum 1.1°C difference among modules. Additionally, it extends battery lifespan by 0.02%, 0.015%, and 0.01% compared to Fuzzy Logic control (FLC), Artificial Neural Network (ANN) and intelligent model predictive control (IMPC), respectively. It also achieves energy savings of 23%, 25% and 15% compared to the FLC, ANN, and IMPC models. Hence, it is evident that the proposed model holds promise for enhancing battery life span with minimal cost in EVs with its simplicity, efficiency, and robustness.