An Investigation of Fault Detection in Electrical Distribution Systems Using Deep Neural Networks

The primary goal of the research is to detect and classify defects in electrical distribution networks using deep learning techniques. At a fault situation, fault voltage, fundamental frequency, and current components are considered for fault identification and categorization. An IEEE 33 bus system is used to model distribution network, and fault conditions are created in simulation to obtain fault components. When a fault occurs current and voltage waveforms contain significant high frequency transient signals. Discrete Wavelet Transform (DWT) and Deep learning (DL) approaches are used to detect and classify the fault in distribution system. DWT is applied for decomposition of high frequency transient signals to extract information in both time and frequency domains. Results show that the proposed Deep Neural Network (DNN) model has high accuracy in recognizing and classifying the fault accordingly. The simulation is done through MATLAB software using deep learning.