Fault Diagnosis and State of Charge Estimation of Battery for Electric Vehicle Applications

The popularity of electric vehicles (EVs) has led to increasing concerns on high-voltage hazards associated with them. It is imperative to carefully examine the method for diagnosing problems in lithium-ion batteries due to their crucial role that led to 30% of EV accidents. The complex nature of faults in EV LIBs makes their treatment challenging. To guarantee the real-time safety of EV operations, quick and efficient diagnosis of battery pack problems is necessary. This study introduces a technique that utilizes Neural Network (NN) models like Multilayer Perceptron (MLP) and Radial Basis Function (RBF) to identify and address EV battery problems. The simulation of the battery for generation of failure-related data is carried out in MATLAB. Enhancing accuracy involves preprocessing the generated data. Later, both models are trained and subjected to testing to assess their performance. The Battery Management System (BMS) is critical in the framework of the LIB system as it helps in maintaining effective and reliable operation by finding the SOC. Hence, the LIB is considered a nonlinear system, and the Extended Kalman Filter (EKF) method plays an effective way of estimating SOC. To increase the reliability and precision of the SOC estimate, an adaptive EKF is adopted based on KF and EKF. In this research, a mathematical model for the BMS is formed by integrating a model-based SOC. estimator. In the proposed work, SOC estimation will be done based on EKF. The suggested system is simulated in MATLAB, and the efficacy of the proposed approach is validated with respect to time.