Deep Learning-Based State-of-Charge Assessment Model for Hybrid Electric Vehicles Energy Management Systems

Abstract

Promoting the use of Electric Vehicles (EVs) is a practical way to encourage carbon impartiality and thwart the environmental problem. Government regulations and user experiences directly correlate with EV batteries and battery management improvements. Alternative engine technologies have become increasingly important in addressing issues with traditional automobiles in recent years. To decarbonize the transportation industry, electric vehicles are practical solutions. It also becomes crucial to switch from conventional to smart homes and from traditional to EV or HEV vehicles. One of the most vital parts of electric vehicles is the battery. When dealing with larger capacity and high-power needs, high-power providing battery packs-which are made up of many batteries-are necessary. These large battery packs are prone to overheating while being charged and discharged, which can lead to a lot of problems. Consequently, it is imperative to employ a battery management system. It is in charge of optimizing the battery pack so that it functions more effectively and safely. This essay's primary goals are to simulate a Battery Management System (BMS) model and examine several approaches to parameter estimation for a battery management system. It also offers suggestions for the BMS's most effective and economical implementation strategies. An efficient battery management system (BMS), primarily used for signaling the battery level of charge, is still a key component among the numerous HEV technologies (SOC). Since excessive charging and discharging always cause damage to the batteries, the BMS must provide an accurate SOC estimation. Although several SOC prediction strategies are available to control battery cell SOC, HEVs require improved SOC estimation capability. The construction of a unique deep learning with SOC estimate model for safe energy management technique for this is the main emphasis of this paper from this perspective. The proposed model uses a hybrid convolution neural network with long short-term memory (HCL) model to precisely estimate SOC. The HCL model is used to facilitate modeling and provides an accurate representation of the input and output association of the battery model. A detailed experimental investigation showed that the proposed model was superior to other current methods in several ways.