This study focuses on enhancing the electrical contact reliability of electric vehicle charging interfaces and constructing a fault prediction model. By analyzing the failure mechanisms of electrical contacts, reliability improvement schemes such as modular wiring harness design and anti-corrosion coating application are proposed. A dynamic monitoring method for contact resistance is established based on IEC standards. A GCN-LSTM deep learning model is constructed to integrate current/voltage time-series data with user behavior features, enabling early fault warning.Experimental validation demonstrates that the model achieves an accuracy of 88.12% and an F1 Score of 0.844 in charging pile fault diagnosis, outperforming traditional models. This research provides theoretical support and technical pathways for intelligent operation and maintenance of charging facilities, which is of great significance for ensuring the healthy development of the new energy vehicle industry.