Empirical Wavelet Transform for Harmonic detection under Dynamic condition

Presence of harmonics - one of the major power quality issues in present day power system is primarily due to the non-linear loads connected, which are inevitable as renewable energy sources become more prominent. Power system, though operates at normal condition for most of the operating time, may be subjected to abnormalities such as an unbalance or faults which arise frequently due to momentary tree contact, lightning strikes, bird/animal contact, and continuous load variation. Despite any aberrant situation, non-linearity of the system continues to draw harmonics from the supply. Therefore, the need for estimation of harmonics at any various situations that may occur in a power system is very important before triggering a suitable compensatory action to solve the power quality issues created by harmonics. This work will investigate the suitability of the Empirical Wavelet Transform (EWT) algorithm in separating the fundamental and harmonic components from the composite current/voltage waveform under various situations that may occur in the power system. This algorithm extracts the different intrinsic mode functions from the input signal by passing it through appropriate wavelet filter bank. For testing the effectiveness of EWT, in this work, a sample power system with thyristor converter fed DC Drive as a non-linear load is considered where source current is a composite waveform containing harmonics. Both synthetic and actual data from this system under various conditions other than normal operation is used for testing the proposed algorithm. Results are presented and discussed.