Analysis Of The Yarn Signal Based On Wavelet Transform

Evenness is the main basis for judge the quality of semi-finished products and yarn, and ismain way to detection the mechanical fault diagnosis of the production process, but is alsoessential to be one of the indicators in the international trade. Spectrum diagram, variationcurve and the yarn defects detection and classification is the main content of the evenness test.For the calculation of the spectrum diagram is traditionally though the Fourier analysismethod, but yarn signal is nonstationary signal, using the wavelet transform may suitable.Wavelet analysis can check up the development trend of the signal. Based on the format ofthe ascension of the second generation wavelets has many merits, chapter5of the thirdquarter, the second generation of the study of adaptive wavelet de-noising, variance and as thebasis for the selection filter squares, make the predictors and scrubbers can always adapt toeach scale the characteristics of approximation signal, and connecting with the soft thresholdand hard threshold of the advantages of the yarn signal and noise，benefit the spectrumdiagram of calculation and other analysis.This paper analysis the wavelet transformation frequency window properties by processingthe yarn signal using wavelet transform, combined with the55bandpass filters, which hasfrequency center and bandwidth while using Fourier method, discusses how to choose thescale when using wavelet transform method to calculate the spectrum. The wavelettransformation of the generating function through the telescopic peace to move signal isanalyzed, the corresponding frequency window will change when scale changed, and waveletfunction in the frequency domain band-pass property and Fourier method to calculate thespectrum in the chart the filter some similar. Through the analysis, and concludes that thescale of the logarithm growth in learning. Using the complex morlet wavelet calculation,compared with Fourier method that the spectrum diagram, and analyzes the reason for thedifference between the produce. The quality factor of the55bandpass filter used in the FFTmethod is approximately equal, and the wavelet transform have the similar properties,therefore using wavelet analysis to calculation of wave spectrum is reasonable. BecauseMorlet complex wavelet spectrum calculation is put in bigger difference, so designed onestrict frequency wavelet function, determined the parameter value, and used to calculate thewave spectrum.