Fault diagnosis of power electronic circuits based on ICA-LSTM

A novel approach, which is based on the Imperial Competition Algorithm (ICA)-optimized Long Short-Term Memory (LSTM) neural network, is proposed to enhance the accuracy of fault diagnosis in power electronic circuits. With circuit faults considered as outputs, rectified waveforms are utilized as inputs. By employing the ICA algorithm to optimize the hyperparameters of the LSTM, a circuit fault diagnosis model based on ICA-LSTM has been successfully developed. The diagnostic performance of the ICA-LSTM model is assessed and contrasted with prevailing diagnostic approaches. The diagnostic accuracy of 98.09% is attained by the ICA-LSTM model, as demonstrated by the results. The effectiveness of the GJO-SVM model in diagnosing faults within power electronic circuits is confirmed through comparative analysis.