Fault Diagnosis and Localization of Power Cables Using Bi-Directional Long Short Term Memory with Adam Optimizer

In recent days, the fault diagnosis and localization of power cables is one of the incredible remote ones, which have to be identified carefully and located, fixed within a short period. In existing, the Support Vector Machine (SVM) is used for the diagnosis of fault in power cables. However, these Machine Learning (ML) models are designed for static data, which struggled to capture the temporal dependencies in a sequential time series signals. Therefore, Bi-directional Long Short Term Memory (BiLSTM) with Adam Optimizer (AO) is proposed and named as BiLSTM-AO. Initially, the procedure is started with the data collection and further processed into preprocessing with Time Domain Reflectometry (TDR). After that, the features are extracted from the preprocessed data using Short Time Fourier Transform (STFT) to extract the apparent features in time domain such as frequency components and harmonics. Finally, the BiLSTM is utilized for the diagnosis of faults and AO is used to increase the performance of model. From the results, the proposed Bi-LSTM-AO achieved better accuracy (99.80%) when compared with existing Whale Gray Wolf Optimization Algorithm-Variational Mode Decomposition-Support Vector Machine (WGWOA-VMD-SVM).