Multi Wavelet Transform with Radial Basis Function Neural Network for Internal Faults Diagnosis in Power Transformer

摘要

In this article, an efficient method is projected for perceiving the internal fault signal of the power transformer. An efficient approach is the amalgamation of multi wavelet transforms (MWT) and radial basis function neural network (RBFNN) method. Initially, the three phase powertransformer behaviors are analyzed under the normal and abnormal conditions. Here, two phases are done, the first one is fault detection process and the second one is fault classification process. For the two processes, primarily, the features of the signal are extracted with the help of multiwavelettransformation. The extracted features are given to the input of RBFNN and the network is qualified by back propagation training algorithm. Then the RBFNN is utilized to categorize the faulty signals of the power transformer. The projected approach is applied in MATLAB/Simulink workingplatform. The performance of the projected method is assessed with six kinds of faulty signals. Performance of the projected method is investigated and its performance associated with those of DWT-RBFNN and multi wavelet-ANN techniques. Then the statistical measures of the proposed and existingmethods are analyzed. From the comparative analysis, it is perceived that the anticipated technique has shown better solutions in the name of accuracy, sensitivity, and specificity.