The Research Of Noise Canncelling Algorithm For Pipeline Leakage Detection Based On DSP

This article analyzes the harm of noise in water leakage localization based on the characteristic of leakage signal and advises two kinds of denoise algorithm: spectrum subtraction and Least Mean Square adaptive algorithm (LMS).Six kinds of spectrum subtraction forms are analyzed and simulated in matlab,finally realized with C language.Program is validated separately in VC++6.0 and CCS platform.Least Mean Square adaptive algorithm (LMS) is a kind of track algorithm, applies it in denoise and simulates it in matlab.Finally, feasibility is validated on DSP using CCS platform.Leakage signal with noise is put as reference.then,validity is contrasted between spectrum subtraction and LMS algorithm. LMS algorithm is chose to denoise in water leakage. The influence of strong noise is removed by "3 time of mean squares" before silencing broad band noise by LMS algorithm,Which is realized on DSP. At the same time, adaptive filter step is adjusted in LMS, time delay may be sure with the position where filter weighting coefficient W appears the extreme value. Accurate rate is higher in matlab simulation, but the capacity of operation is big. It does not suit the realtime processing , further improvement is needed in practice.Finally, paper is summarized and some advice is given in future works.