Method Based on Improved Fastica-lsm for Harmonic Detection in Power System

This paper proposes a new method based on improved FastICA (fast independent component analysis)LSM (the least square method), in order to improve the accuracy and speed of harmonic detection in power system. This work ameliorates the traditional FastICA algorithm by defining the initial value of demixing matrix iteration, thus inducing the iteration times and eliminating the problem of higher-order convergence FastICA sensitive to initial value, and harmonics and fundamental components are extracted from electric signals using fifth-order convergence FastICA. Finally LSM is used to calculate the amplitude and phase angle of separation components. Simulated and real signals are used to test the method. Results shows that the proposed method has faster detection rate and higher accuracy compared with other methods available in the literature.