Study on Denoising Yarn Tension Signal Based on Blind Source Separation
To address the problem of low accuracy of the collected yarn tension signal and the difficulty in reading the tension value,a blind source separation method of the yarn tension signal combined with empirical modal decomposition(EMD),singular value decomposition(SVD)and fast independent component analysis(FastICA)was proposed.The empirical modal decomposition method was applied to adaptive decomposition of tension signal to obtain intrinsic modal function(IMF)components which have smooth and linear characteristics.The intrinsic modal function and the tension signal were formed into a multidimensional observation signal,and covariance matrix was decomposed by singular value decomposition to calculate the adjacent singular value differences and determine the number of source signals.The correlation coefficients between the IMF components and the tension signals were calculated,and the IMF components were selected to be reconstructed with the tension signals to obtain new multichannel signals.Fast independent component analysis was performed on the obtained multichannel observation signals to achieve noise separation of the yarn tension signals.The experimental platform denoising experiment was built to verify the algorithm.The results show that the method can effectively separate the yarn tension signal and improve the signal-to-noise ratio.The signal-to-noise ratio is improved by 2.678 1 dB compared with the 15-layer wavelet decomposition denoising method,which completes the noise removal of the yarn tension free vibration signal.
国家科技期刊平台
NSTL
万方数据
