Identification of the time-varying pattern of periodic harmonics

With the widespread use of power electronics equipment and other nonlinear loads since the last twenty years, current and voltage waveforms are becoming highly distorted. Considering the effects of harmonics on the performance and life expectancy of electrical devices, prediction of harmonic levels is important to prevent the risks of their disturbances and to optimize the size of equipment. For this purpose, two estimation tools have been developed and compared to model time evolution of periodic harmonics. The first one is based on a discrete Fourier transform; the second one uses a stochastic approach based on a continuous-time model. The harmonic phenomenon is then considered as a time-dependent stochastic process. The two models have been applied to the identification of the harmonic currents drawn by an institutional building owning a weekly load pattern. The results have been then compared to get conclusion on the most appropriate method in the characterization of the time varying behavior of harmonics. The long term objective is to use these models to simulate multi-converters installations and to predict the effects of harmonics on electrical equipment such as transformers´ overheating troubles.