Empirical mode decomposition with Hilbert transform for classification of voltage sag causes using probabilistic neural network

This paper employs Empirical Mode Decomposition (EMD) combined with Hilbert Transform (HT) to detect the voltage sag causes. Any power quality disturbance waveform can be seen as superimposition of various oscillating modes. It becomes necessary to separate different components of single frequency or narrow band of frequencies from a non-stationary signal to identify the causes which contribute to power quality disturbances. The main characteristic feature of EMD is that it decomposes a non-stationary signal into mono component and symmetric signals called Intrinsic Mode Functions (IMFs). Further, the Hilbert transform is applied to each IMF to extract the features. Then, Probabilistic Neural Network (PNN) classifier is constructed based on EMD which classifies these extracted features to identify the type of voltage sag cause. Three voltage sag causes are taken for classification (i) fault induced voltage sag, (ii) starting of induction motor and (iii) three phase transformer energization. A comparison of EMD with Wavelet Transform (WT) is made. The performance of PNN is compared with Multilayer Neural Network (MLNN) based on the above mentioned two methods. Simulation results show that the EMD method in combination with PNN is more efficient in classifying the voltage sag causes.

► Empirical mode decomposition along with Hilbert transform to extract the features.
► Does non-constrained decomposition and gives an idea of Instantaneous frequency.
► Decomposes disturbance signal into monocomponent and symmetric signals called IMFs.
► Hilbert transform is used to calculate instantaneous amplitude and phase of IMFs.
► Classification of voltage sag causes is performed with PNN and MLNN.