Effect Of Fiber Fineness And Length Distribution On Irregularity Of The Yarn And The Predication Model Of Yarn Qualities

Yarn unevenness is consist of the random distribution irregularity (or named the theoretical limit irregularity) and the additional irregularity caused by mechanical factors and spinning process. Generally, when irregularity theories are discussed, yarn random distribution irregularity is more concerned while the additional irregularity is ignored. However, the actual unevenness, combination of the two irregularities' effects, is more concerned in practical application.This paper analyzes the relationship among fiber fineness, fineness irregularity and theoretical yarn irregularity. It is concluded that fiber fineness irregularity has a certain effect on yarn evenness, which is less than that of fineness itself. Therefore, fiber fineness is the major reason of the ultimate yarn limit irregularity. Meanwhile, the relationship between fiber length distribution and theoretical yarn irregularity is further analyzed and summarized in this paper, referring to some classical algorithms on foreign literatures. Based on the ideal model of fiber-length-array, comprehensive effects of fiber fineness and fiber length on theoretical yarn irregularity are discussed in the paper. This paper assumes the fiber length density distribution function for the Triangle-shaped distribution f(x). Some fiber length statistical values such as the average fiber length (ML), the upper quartile length (UQL) and short fiber content (SFC) are established due to the characteristics of f(x). Then the theoretical irregularity is derived by using the above statistical values. From the formula, the relationship among fiber length, fineness and the theoretical unevenness is obtained. After simplifying the equation and analyzing the relationship among fiber fineness, fiber length and theoretical yarn irregularity, it is showed that the thinner fiber fineness and the longer fiber length lead to the lower theoretical yarn irregularity.Besides the above research, this paper contains some prediction work towards yarn quality. Due to the multi-process of yarn production and complicated physical structure of a yarn, there is a complex nonlinear relationship among the fiber quality, the parameters of the spinning and the yarn quality in the process of spinning. It is difficult to describe the relationship clearly in an ordinary mathematical model. So it is necessary and meaningful to find a more appropriate way to describe the relationship. In the paper, a combination of principal component analysis and back-propagation neural network model is applied to predict yarn quality. The basic idea is to simplify the input grid by use of principal component analysis to compress the various tested values of fiber into several new variables. This approach makes the training speed of the grid faster and the output more accurate. Finally, the precision forecasting model is matched with the historical data of factory, and the speed of the convergence rate is quicker, the residual errors is lower, and the average error is lower than 7.0%. So it is proved that the predicting model is suitable for use.