Harmonic detection using fuzzy LMS algorithm for active power filter

Successful control of active power filter requires an accurate current reference. However the conventional detecting methods for harmonic currents have various limitations. Base on adaptive interference canceling theory, this paper presents an adaptive least mean squares (LMS) algorithm employing fuzzy step size for the fast estimation of harmonic current signals in power networks. The learning parameter of the proposed algorithm is constrained by two variable parameters. By using a fuzzy logic based step size update method which causes an automatic suitable adjustment of the step size, the present method can provide fast convergence and noise rejection for the tracking of fundamental and harmonic components from distorted signals. The results of computer simulations and physical experiments have proved the proposed algorithm is practicable.