Detection and classification of power system fault in IEEE 30 bus network using wavelet transform and novel hybrid Bees Bayesian Optimization algorithm based Improved convolution Neural network (ICNN)

This work suggests the fault detection and classification technique in an IEEE standard 30 bus network using Wavelet Transform (WT) and Bees Optimization Algorithm (BOA) based Improved Convolution Neural Network (ICNN). The post-fault the voltage and current signals were collected from the Bus via WT, and the proposed model initially identifies important statistical parameters from one cycle of those signals. The WT is a signal analysis algorithm which decomposes power signals into multiple frequency ranges using a series of high-pass and low-pass filters. The extracted features are normalized and supplied as input to ICNN which is used for classification of fault in the system. The ICNN procedure begins with convolution, max pooling and feature analysis, and the final step is feeding the results to the fully connected (FC) layer. A novel hybrid Bees Bayesian Optimization Algorithm (BA-BO) is employed for the tuning of parameters in ICNN. The simulation of IEEE 30 bus system is done by using MATLAB Simulink and then the results were discussed. The proposed technique provides an overall testing accuracy of 99.67% compared to existing methods.