부산대학교 도서관

This paper describes a method of cancelling the periodic noise induced from the power-line during measurement, using the LMS algorithm and the adaptive periodic noise canceller synchronized to the power-line frequency. To simplify the analysis, a modelling is made whereby the signal is a random variable and the noise is periodic, containing fundamental and a large number of harmonic components with the power-line frequency as the fundamental component. A parameter for evaluating the noise cancelling performance of the adaptive periodic noise canceller is derived analytically. It is shown that the processing can be performed without multiplication, without degrading the performance. Then a fast algorithm is proposed for adaptive periodic noise cancellation, which can reduce drastically the computational complexity, thereby degrading the performance little. By applying the algorithm, the real-time processing by a microprocessor is realized with a sampling rate of several kilo- hertz. Finally, a discussion is made of the effect of digital implementation on the per- formance, and it is shown that the cancellation of the power-line interference by the fast algorithm is practically useful. [ABSTRACT FROM AUTHOR]