Comparison of Power System Harmonic Prediction

Due to the advance in technology nowadays the application of nonlinear load is very demanding. The nonlinear loads that consist of power electronics devices in many applications may leads to power quality problems, and one of the major problems in power quality is harmonics. Active filter is a common tool has been used for harmonic mitigation, however the active filter use digital signal processing (DSP) to calculate the harmonic information which cause time delay between measured and injected harmonic that can results in failure to fully cancel the harmonics. This paper presents the concept, method and function of adaptive filters using Least Mean Square (LMS) and Recursive Least Square (RLS) to predict real time harmonics that may use in active filter to compensate for the time delay caused by the process of acquiring harmonic information. The results shown that the RLS and NLMS algorithms present a better performance than LMS and the mean square error (MSE) in general is increased when the harmonic order numbers that construct the signal increases.