Discrimination of Power Quality Distorted Signals Based on Time-frequency Analysis and Probabilistic Neural Network

Recognition and classification of Power Quality Distorted Signals (PQDSs) in power systems is an essential duty. One of the noteworthy issues in Power Quality Analysis (PQA) is identification of distorted signals using an efficient scheme. This paper recommends a Time–Frequency Analysis (TFA), for extracting features, so-called "hybrid approach", using incorporation of Multi Resolution Analysis (MRA) and Generalized S Transform (GST). Moreover, the proposed scheme is noticed to quality of features and ranking them in order to find the best combination with lower dimension. A new efficient feature ranking method namely Orthogonal Forward Selection (OFS) is applied for selection of the best subset features. Probabilistic Neural Network (PNN) as classifier is considered. An extensive series of simple and complex PQDSs are simulated to verify of suggested detection scheme. Also, sensitivity of the proposed method under different conditions of noise has been investigated. The obtained outcomes are compared with those obtained using other methods in previous research to assess the performance.