Using the Principal Component Analysis and BP Network to Model the Worsted Fore-spinning Working Procedure

The characteristic of worsted fore-spinning working procedure and the BP neural network modeling technology all have been summarily analyzed. Based on roving craft parameters´ relevant characteristic, the principal component analysis method is proposed to pretreat the sample data set, which results are the new sample data of the BP neural network. The input layer node numbers reduce; the relevancies among every input factors are eliminated simultaneously; the network architecture is simplified that the network´s study speed and the performance are all enhanced greatly. The relative mean error percent (MEP) between the roving unevenness and weight´s prediction value and their mean value, reduce to 3.35% and 1.95% respectively. While the former values respectively are 4.86% and 2.24% The network precision, the correlation coefficient between the forecast value and the actual value all have the remarkable enhancement.