This study introduces an AI-powered digital twin solution for intelligent fault detection and predictive repair in electric vehicles (EVs) to meet the increased demand for reliability, safety, and cost-effectiveness in next-generation transportation systems. The system combines real-time sensor readings such as battery temperature, motor phase current, and vibration amplitude gathered from virtual prototypes and hardware simulations and uses them to monitor the health condition of important EV components in real-time. 3200 samples of data were obtained, preprocessed, and utilized for training four machine learning models: Random Forest, Support Vector Machine (SVM), XGBoost, and Long Short-Term Memory (LSTM). They were tested for their capability to identify early faults and estimate Remaining Useful Life (RUL) of parts with precision. Out of them, the LSTM model had the best accuracy, precision, and reliability and was thus best suited for real-time prediction applications. The MATLAB/Simulink simulated digital twin provided end-to-end visualization of load thermal behavior, motor current, and mechanical vibration, confirming the predictive findings established by the Artificial Intelligence (AI) models. Fault notifications were initiated when parameters crossed thresholds learned during training, allowing for timely maintenance decisions. Confusion matrices and performance measurements also attested to the effectiveness of each model in identifying degradation trends. Combining digital twin technology with AI enables predictive maintenance, lowering unplanned breakdowns, increasing vehicle availability, and prolonging the lifespan of components. There are robust practical implications of this research in fleet management systems, EV production, and autonomous vehicle monitoring systems, providing a scalable, intelligent, and powerful maintenance solution for next-generation electric mobility platforms.