Classification of Power Quality Disturbances using Adaptive Variational Mode Decomposition based Random Vector Functional Link Network

This paper presents power quality (PQ) disturbance detection and classification using Adaptive Variational Mode Decomposition (AVMD) based Random Vector Functional Link Network (RVFLN), which has been introduced as a new contribution to earlier studies. In VMD technique, it is very difficult to select the mode number and the value of penalty parameter. That's why both the values are optimized using firefly algorithm. AV-MD is an improved version of VMD technique. Various power quality disturbances are generated synthetically and the non-stationary voltage signal samples are extracted. These extracted waveforms are subjected to AVMD algorithm. Distinguishable features are extracted from the output of the modes of AVINID. Finally, all the extracted features are classified using RVFLN which has been proved to he a very powerful classifier. Applicability of the proposed scheme with RVFLN has been tested for different noisy conditions. A comparative study has been done with the other established techniques such as Empirical Mode Decomposition (EMD) and Empirical Wavelet Transform (EWT) where it has been established that the proposed technique gives better performance. Classification accuracy achieved with the proposed technique is satisfactory and acceptable.