Neural network-based detection and recognition method for power quality disturbances signal

With widespread use of various kinds of electric devices, the demand for clean power supply has been increasing in the past decades, which has great effect on the energy market. In order to improve supply quality of power system, the sources and causes of power quality disturbances must be known before appropriate mitigating operation. This paper used wavelet network-based neural classifier to automatically detect, localize, and classify the transient disturbance pattern, which can acquire the qualitative and quantitative results. The wavelet transform can offer a better compromise in terms of time-frequency domain, which decomposes the transient signal into a series of wavelet coefficients, corresponding to a specific octave frequency band containing more detailed information. To acquire the original information of transient signal, the wavelet-based denoising technology is discussed in a low signal noise ratio environment. The improved training algorithm is utilized to complete the neural network parameters initialization and classification performance. In order to satisfy power system observation, the power quality monitor configuration method is proposed. The testing results and analysis indicate that the proposed method is feasible and practical for analyzing power quality disturbances.