XAI-OptiEV: Explainable DL Framework for Optical Fiber Fault Detection to Optimize Energy Efficiency at EV Charging Stations

The swift growth of smart technology is reshaping the existing paradigm of energy management and fault detection in electric vehicle (EV) charging ecosystems to further interconnect systems with optimized efficiency in EV charging stations and optical fibre networks. This paper focuses on XAIOptiEV, an innovative framework that combines optimizing energy consumption and the detection of communication infrastructure faults in EV charging stations. Using various deep learning (DL) models, including Feedforward Neural Networks (FFNN), 1D Convolutional Neural Networks (1D CNN), and Recurrent Neural Networks (RNN), we statistically analyze the model performances across different metrics, such as accuracy, precision, recall, F1-score, and loss. Our findings suggest that the RNN model surpasses the rest of the model’s ability to correctly detect faults in optical fibers. Moreover, we apply XAI methods, namely SHAP and LIME, to deliver intuitively graspable information regarding model decisions, elevating trust and transparency. This integrated approach by means of DL with XAI is a major factor in the enhancement of the reliability of energy delivery and the support of sustainable activities in the EV charging ecosystem.