Noise separation of the yarn tension signal on twister using FastICA

Some unknown noise will influence the wave appearance of the twist yarn tension signal in different factory environments or production processes. It is difficult to analyse the tension signal pattern and recognize unusual defects by the on-line yarn quality control system. In this paper, the independent component analysis (FastICA: fixed-point independent component analysis) is applied to separate the unknown noise signal from the unusual tension signal on the yarn twist machine. Different from the traditional low-pass filter (e.g., Butterworth filter), FastICA can not only successfully separate the noise with different types but also remain the main tension signal information. Firstly, FastICA algorithm is introduced, and then simulation experiments and on-line tests are carried out to evaluate the performance of this method and traditional low-pass filter.