JOURNAL OF TEXTILE RESEARCH ›› 2017, Vol. 38 ›› Issue (07): 90-94.doi: 10.13475/j.fzxb.20161003405

Previous Articles     Next Articles

Method of mixed dye liquor concentration detection based on spectral optimization and Multi-level support vector machine

  

  • Received:2016-10-14 Revised:2017-03-31 Online:2017-07-15 Published:2017-07-18

Abstract:

In order to improve the detection accuracy of multi-components mixed dye liquor concentration during printing & dyeing process, a method was proposed for simultaneous determination of three components dye liquor concentration by absorption spectroscopy. The absorption spectra of the multi-components mixed dye liquor was collected by USB2000+ microfiber spectrometer with wavelength range of 339.22~1 000 nm. The original effective absorption spectra were smoothed by Savitzky-Golay (SG) convolution smoothing. 80 mixed dye liquors were divided into 75 training groups and 5 test groups according to sample set partitioning on joint X-Y distance (SPXY) method. The optimal characteristic wavelengths of the successive projections algorithm are 22. The multi-component concentration of the dye solution was detected by using the multi-level multi support vector machine model. The experimental results show that the relative error of the three-components mixed dye liquor concentration can be controlled within 10%, and the training time and prediction time of the classification model are shortened by about a half, which provides a reference to realiae on-line concentration detection of multi-component dye in the later stage.

Key words: successive profections algorithm, multi-level support vector, characteristec wavelength, multi-component dye solution, concentration detection

No related articles found!
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!