Power Quality Data Mining Using Improved Probabilistic Neural Network

In this paper, we propose a new technique to enhance the learning capabilities and reduce the computation intensity of probabilistic neural network using the kmeans clustering algorithm for recognizing and classifying power quality disturbances. First, the Discrete Wavelet Transform is employed to extract the energy distribution features of the distorted signal at different resolution levels. Then, kmeans algorithm is applied to the training dataset to reduce the amount of samples to be presented to the neural network, by automatically selecting an optimal set of samples. Finally, the Probabilistic Neural Network classifies these extracted features to identify the disturbance type according to the energy features and disturbance duration. The obtained results demonstrate that the proposed technique performs exceptionally in terms of both accuracy and computation time when applied to the power quality disturbances data set compared to a standard learning schema that uses the full data set. The method is described and demonstrated using 6 different sets of power disturbances.