Neural network-based detection and localization of short duration disturbances in power system

The power quality issues have become an exponentially demanding research field for electric utilities and customers, which usually involve transient variation in power supply voltage or current. This paper presents a new method using wavelet transformation and neural network, investigating the power quality in qualitative and quantitative results. The major feature of wavelet transformation is the multi-resolution analysis technique which can decompose the transient signal into several signals with different levels of resolution. From the decomposed signals, the original signal can be recovered without losing any information. By means of signal singularity detection analysis, the wavelet transformation can accurately detect and locate the transient voltage waveform. The combination of wavelet transformation with neural network has make progress in neural network, where the wavelet is introduced as activation function of the hidden neurons with a linear output neuron. The simulation results illustrate the efficiency of the proposed method in power quality disturbances detection and analysis.