Leakage Diagnosis for Pipelines Based on ICA and Wave Structure Recognition
Wei Liang · Huagong zidonghua ji yibiao · 2007
For solving the difficulty of feature signal extraction with strong noise in negative pressure wave,a new method based on Independent Component Analysis(ICA) is proposed to realize separation and filtering between weak aim signal and strong noise.The experiment shows that signal-to-noise ratio of processed pressure signals is raised obviously.The pressure wave in different operation modes are analyzed and the wave structure recognition is proposed to judge pipeline condition.The characteristic wave vectors are selected and then the mode base of pipeline conditions is established in limit samples.The testing results show the reliability and validity of this method.