Micro-Grid Fault Detection and Classification for Smart Vehicles Using Simple and Efficient Metapath Aggregated Network-Sea Horse Optimization Approach

Microgrid (MG) Fault Detection (FD) and classification are crucial for maintaining stability and reliability, especially with Smart Vehicle (SV) integration. Effective FD and classification helps identify and mitigate faults quickly, preventing power disruptions and ensuring smooth operation of the system. The system struggles with bi-directional power flow since identifying normal fluctuations from faults becomes a problem resulting in incorrect assessments or delayed operations. To overcome these drawbacks, this paper proposes an improving MG FD and Classification for SV using Simple and Efficient Metapath Aggregated Network (SEMAN) and Sea Horse Optimizer (SHO) approach. The process begins by gathering data from operational MG infrastructure dataset, which is then passed through a preprocessing phase. Fairness-Aware collaborative filtering (FACF) are employed to clean and remove the missing value in the input data. Once pre-processed, the data enters the prediction and classification phase, to enhance the accuracy of predictions. The phase-to-ground, phase-to-phase, phase-phase-to-ground and three phase faults are successfully predicted and classified by using SEMAN. The weight parameter of SEMAN is optimized using SHO. The SEMAN-SHO technique is implemented in MATLAB and evaluated using various performance metrics, including accuracy, precision, recall, and Root Mean Squared Error (RMSE). The results demonstrates that the SEMAN-SHO method performs better than the existing approaches, such as Zebra Optimization Algorithm with Spiking Neural Network (ZOA-SNN), Long Short-Term Memory- Adaptive Neuro Fuzzy Inference System (LSTM-ANFIS), Machine Learning (ML), Particle Swarm Optimization Back Propagation (PSO-BP) and Convolutional Neural Network- Long Short-Term Memory (CNN-LSTM). These results demonstrate the effectiveness of the proposed method in accurately detecting and classifying faults in MG for SV, with a high accuracy of 98.5%, precision of 98.4%, recall of 98.9%, and an RMSE of 0.62, ensuring reliable and stable system operation.