Adaptive Harmonic Detection Algorithm Based On New Variable Step-size Least Mean Square

With the rapid development of power electronics technology, equipment produces a large number of harmonic during transforming electrical energy, which gives a serious impact on the performance of the power system. A variety of precision equipments and instruments have a high demand on the quality of power supply. It is necessary for detecting and treating harmonic as these reasons.This paper briefly introduces the harmonic concepts, harm and the current situation of harmonic treatment. It also introduces the relevant theory of harmonic detection and methods of harmonic detection in detail. Then it compares these methods. As the research on Least Mean Square (LMS) of the adaptive algorithm, the paper proposes new adaptive algorithms of harmonic detection. Fixed step size largely affects LMS, and there is a contradictory relationship between convergence speed and steady-state error. So, variable step size LMS algorithm is proposed and used in adaptive harmonic detection algorithm.On this basis, this paper presents new variable step size LMS based on adaptive harmonic detection algorithm.First, hyperbolic cosine function is applied to the function of variable step size. Then improve the hyperbolic cosine function. The curve of the improved hyperbolic cosine function would slow down the rate of change when it is close to the zero of coordinate. This characteristic can reduce the step size in the late stage of convergence to reduce the algorithm’s steady-state error.Second, combine the variable step size LMS algorithm based on the Rayleigh distribution with harmonic detection technology. The change of Rayleigh distribution function curve is rich, increasing step size in the beginning of convergence and reducing step size in the late stage of convergence. Then, it can reduce the algorithm of the steady-state error.Finally, the characteristic that the higher-order cumulants have is inhibiting white Gaussian noise. This characteristic can be applied to the function of variable step size to get the HOC Variable Step Size LMS algorithm. The algorithm combines with the algorithm of the adaptive harmonic detection. It also improves the error function. The instantaneous error value is replaced by error accumulation. So the anti-noise performance of the algorithm is improved.The proposed algorithm has small mean square error and better harmonic detection results. The computer simulation tests the effectiveness of the algorithm.