Prediction of Electric Vehicle Battery State using SHAP Model | IJCT Volume 13 – Issue 1 | IJCT-V13I1P18

The Battery Health Prediction and Defect Detection System is an intelligent and scalable method of proactive battery health maintenance. It uses machine learning algorithms and deep learning models to make predictions about the State of Health (SoH) of batteries and make predictions about potential degradation. The system uses machine learning algorithms such as Random Forest, LSTM, SVM, and Decision Trees to make predictions based on critical parameters of battery health, including voltage, current, temperature, internal resistance, and cycle counts. The decision-making process is made clearer and more understandable by using SHAP analysis. The system is built as a web application utilizing the Django framework and backed by a MySQL database. The system allows secure user interaction, real-time predictions, model handling, and viewing past predictions. The system has been shown to be accurate in experimental validation conducted by utilizing metrics of MAE, Root Mean Squared Error, and R², thus ensuring that it is effective in electric vehicle batteries, renewable energy storage solutions, and industrial automation, thus ensuring predictive maintenance of batteries. Future enhancements focus on making the system more intelligent, scalable, and adaptive for real-world battery monitoring. IoT integration and expanded sensor parameters will enable real-time data collection, instant health predictions, and predictive maintenance. Advanced deep learning models, including transformers and hybrid architectures, will improve accuracy in forecasting battery degradation. Mobile applications, cloud–edge computing, and federated learning will enhance accessibility, scalability, and data security. Improved explainable AI and adaptive learning will increase transparency, trust, and continuous performance improvement.