A new morphological denoising algorithm to remove the noise of power signals

There exists noise in measurements of power signals. We studied structuring elements (SE) and their influence to denoising properties, and proposed a new denoising algorithm. The new algorithm uses the median morphological operation including the median opening-closing operation and the median closing-opening operation to improve denoising properties, and uses the cosine SE and the circle SE to make up the double structuring elements. In order to speed up convergence and decrease the steady state error, we take the least mean square (LMS) error as a cost function and introduce an input signal factor. Thus, we establish a variable step-size LMS (VSSLMS) adaptive algorithm. Experimental results show that the new algorithm can quickly trace changes of the system, adaptively regulate convergence step-size. It can effectively remove the noise of power signals, and has a good application prospect in signal denoising.