Optimization of Prediction Performance for Worsted Yarn Based on Neural Network

摘要：

One-hidden layer and two-hidden layer Backpropagation neural network(BP) models were used to predict both unevenness value (CV) and breaking strength (BS) of worsted yarn under the conditions of large-scale input samples,high input dimensions and the number of hidden layers nodes were of 6,7,8,respectively. And the optimum model in predicting CV or BS was selected to make a comparison with Radial basis function network (RBF) model. The results showed that one-hidden layer BP neural network with 8 hidden layer nodes was the most suitable in the prediction of CV,while two-hidden layer BP neural network with 7 hidden layer nodes was better than others in predicting BS. In addition,both BP neural network models were superior to RBF neural network models in predicting performances.

关键词： Worsted yarn One-hidden layer Two-hidden layer Back-Propagation Radial Basis Function

作者: Xiang Li Zhiqin Peng Zongdong Gu Yuan Xue Guoliang Hu

作者单位: College of Materials and Textiles,Zhejiang Sci-Tech University,Hangzhou 310018,China Zhejiang Linglong Textile Co.,Ltd.,Jiashan,Zhejiang 314104,China College of Garment and Art Design,Jiaxing University,Jiaxing,Zhejiang 314001,China

会议类型: 国际会议

会议名称: The Second International Conference on Advanced Textile Materials & Manufacturing Technology(第二届先进纺织材料及加工技术国际会议)

会议地点: 杭州

会议语种:英文

页码: 516-520

在线出版日期: 2010-10-20（万方平台首次上网日期，不代表论文的发表时间）