An Explainable Spatio-Temporal Deep Learning Framework (ST- XFNet) for IoT Sensor Fault Detection and Classification in Electrical Systems

AbstractReliable fault detection and classification (FDC) is critical for the safe and uninterruptedoperation of modern electrical and Internet-of-Things (IoT) enabled industrial systems.Traditional rule-based and signal-processing techniques struggle with the high-dimensional,nonlinear and noisy data streams produced by distributed sensor networks, whilecontemporary deep learning models, although accurate, suffer from limited interpretability,sensitivity to noise and class imbalance. This paper proposes ST-XFNet, an explainablespatio-temporal deep learning framework that integrates a one-dimensional ConvolutionalNeural Network (1D-CNN) for spatial feature extraction, a Bidirectional Long Short-TermMemory (BiLSTM) network for temporal dependency learning, an attention mechanism foradaptive feature weighting and a SHapley Additive exPlanations (SHAP) module for post-hocinterpretation. Synthetic bias, drift, random and polynomial-drift faults are injected into theIntel Berkeley Research Lab IoT sensor dataset to construct a balanced multi-classbenchmark. Extensive experiments demonstrate that the proposed model attains aclassification accuracy of approximately 99%, an ROC-AUC of 0.996 and maintains over97% accuracy under additive Gaussian noise, outperforming standalone CNN, LSTM andCNN-LSTM baselines. SHAP analysis further reveals that temperature and voltage are themost influential features for fault discrimination, providing actionable insight formaintenance engineers. The results confirm that ST-XFNet offers an accurate, robust andinterpretable solution for proactive fault diagnosis in smart electrical infrastructures.