A New Concept of an Intelligent Protection System Based on a Discrete Wavelet Transform and Neural Network Method for Smart Grids

Because of advances in technology and increased pollution problems, most countries in the world use renewable energy in power generation. To avoid power outages, utility companies obligation to identify and locate the main causes of faults as soon as possible to protect energy systems. In this paper, a new technique for fault classification and detection in the transmission lines of micro-grids using a Discrete Wavelet Transform (DWT) and a Back-Propagation Neural Network (BPNN) is proposed. MATLAB is used to complete the simulation and training process of the neural network. The Daubechies4 mother wavelet 'Db4' is used to decompose the high-frequency components of these signals. Wavelet Transform Coefficients (WTCs) and Wavelet Energy Coefficients (WPCs) are used to classify faults and detect patterns that are used as inputs for back propagation in neural network training. This information is then fed into the neural network to classify and detect the fault. This paper proposes a Wavelet Transform (WT)-based fault-detection method and disturbance-recognition method. To detect the faults, voltage signals are collected under fault conditions and processed by WT. The simulation also shows that the new algorithm is reliable and accurate.