Correlation Analysis Between Uster Hairiness Index And The Number Of Hairs About Cotton-type Yarn

Yarn hairiness is the protruding end of the staple yarn, meanwhile, it may exist in different farms, including fibre ends, fibre loops, and wild fibres. High hairiness influences the processing of yarn in the down stream processing and the property of fabric. Yarn hairiness plays an important role in determing the quality of yarn.At present, there are two methods to measuring the hairiness, projection count method and photoelectric measure method. The classical meter of the projection count method is YG172B and YG173. The measuring result is the hairs number of the certain hair length. It is used to adjust the roving twist multiplier, the parameter of spinning process and the speed of winding, and analyze the reason of warp breakage in weaving and so on. The representative meter of the photoelectric measure method is the module of measuring hairiness in Uster3tester. The measuring result is the hairiness index (H) which is the total length of the protruding fibres with reference to the sensing length of lcm. H is served the business at national and international by refering to Uster Statistics. In order to make the conversion between the number of hair and the standard of yarn hairiness, this study aims at establishing a model that should be able to describe the relationgship between the number of the hairiness index.Seventy-two yarn samples, made of various types of pure yarns and combination yarns of varying count, of different twist and the type of spinning system were collected form13mills. Yarn hairiness was evaluated by YG173hairiness meter and Uster-3tester, the number of1,2,3,4,5,7,10and12mm hairs and the hairiness index (H) were calculated. Using YG173measures the number of hairs different length group and Uster-3measures the total length of hairs, the following conclusion has been found out:there is a very good correlation between type of spinning system and the scatter of hair length; the yarns that hairiness indexs are near have different number of hair, the hair number of yarn made form ring spinning is much more than from no ring spinning. In conclusion, all yarns should be separate into ring spinning and no ring spinning, at the same time, are respective analysed by the established model.Correlation analysis was performed between the single variable and Uster hairiness index. In the sample of ring spinning, the variable which has high correlation with H is twist, the number of lmm hairs, the weighted sum of these numbers of various length hairs (termed K) and linear density. In the sample of no ring spinning, the variable which has high correlation with H is the number of lmm hairs square, the number of lmm hairs cube, the number of lmm hairs and K. The result showed that no direct distinct statistical relationship could be evidenced on respective comparing twist factor, strength, elongation at break and hairiness index in all yarns.Consequently, the independent variables at multiple regression should get rid of twist factor, strength, elongation at break.Through the multiple regressions find a great model for ring spinning. Through the t and F test, every dependent variable is significant in the0.05confidence level. The correlation coefficient of model is equal to0.968. The expression of model is as follow: H=0.020×linear density-0.243×the number of lmm hairs+0.057×K+1.414. And, for the no ring spinning yarn, through the t and F test, every dependent variable is significant in the0.05confidence level. The correlation coefficient of model is equal to0.963. The expression of no ring spinning model is as follow:H=0.036×linear density-0.006×twist+0.0000171×the number of lmm hairs square+0.037×the number of3mm hairs+3.162. There are thirty-four kinds of all yarns which are able to research their actual level of yarn hairiness. According to verifying two models, the study finds out the number of yarn that the actual level of yarn hairiness is different form itself forecast level are equal to6.According to using the BP and RBF network to training and verifying the experimental date, the study obtains the best model for yarn of ring spinning, and the model is as follow: the number of BP net layer are equal to2, the transmission functions are tansig and purelin., the hidden nodes are equal to19, the training function is trainlm. The best model for yarn of no ring spinning is as follow:the number of BP net layer are equal to2, the transmission functions are tansig and purelin, the hidden nodes are equal to16, the training function is trainlm. The verified result is that the forecast level of yarn hairiness is the same to the actual level of yarn hairiness. So, the BP net model is able to exact forecast H. Radial basis function network model has the high training speed, however, the accuracy predicting the Uster hairiness index is bad.Comparing with the multiple regression and RBF network, the forecast of BP net is the best. In a word, the hairiness index can be exact forecasted with the number of lmm hairs by BP net model. As the number of yarn type accuracy of model should be higher.