Research On The Classification Of Long-staple Cotton Based On Support Vector Machine

Nowadays, most of the textile mills are based on the commercial gradeclassification of long-staple cotton in artificial method, which caused some issuessuch as data deviation, less amount of information, personal preferences anddifference in standards between different textile mills, these issues are not benefit tomake the technological progress and stabilize the production quality. With thepromotion of the instrument testing and HVI, a urgent problem in the current is howto classify the long-staple cotton according to the test results of HVI. In order to carryout scientific and reasonable classification of long-staple cotton, accurate indicators oflong-staple cotton and classification of scientific and reasonable are crucial.This paper selects five major internal indicators of long-staple cotton’s fibrequality, which is obtained from the use of partial correlation analysis with detectedindicators of HVI data of the long-staple cotton. The five major internal indicators areLength Uniformity, Fibre Strength, Micronaire, Upper half mean length and Neps,which are the data basis of the classification model of long-staple cotton. For the traitof small sample size and too much data indicators on classification of cotton,one-against-one and one-against-all support vector machine classification models areestablished by using support vector machine categorization algorithm to research theclassification of inverntory menkwa cotton in a textile mill in Xinjiang, and get theclassification results. The result is better than using commercial grade classification,better reflecting the performance of cotton spinning, which can guide the distributionof cotton, and ensure production continuity and stability of the spinning quality.