The ANN virtual manufacturing model of worsted yarn
The sub-ANN predicting models for yarn quality and performance, i. e. yarn unevenness, thin places, thick places, yarn neps, hairiness, ends-down per 1000 spindle-hour, yarn strength and extension at break are built respectively based on artificial neural network (ANN). Through these predicting models, the virtual manufacturing of worsted yarn is exploringly researched and discussed. And the mean relative error of all models doesnt exceed 6％ according to the experiential and theoretic analysis. Except for hairiness and strength, the square of correlation coefficients, R<2>, of thick places per kilometre, thin places per kilometre, yarn neps, yam unevenness, ends-down and extension at break are all higher than 0. 90. The results of the fiber diameter influencing on the spinning performance and yarn properties using ANN models show that, with the increase of the diameter, thick places, hairiness, unevenness and end-down are all increased while thin places, yam neps, strength and extension at break are reduce.After the top neps of 3<＃> sample was adjusted, hairiness, extension at break and ends-down can be improved.
ANN virtual manufacturing model worsted yarn
Yin Xianggang Xiang Qian Liu Gui Yu Weidong
College of Textiles, Donghua University, Shanghai 200051,China Wuxi Entry-Exit Inspection & Quarant College of Textiles, Donghua University, Shanghai 200051,China College of Textiles, Donghua University, Shanghai 200051,China Wuhan University of Science and Engin
国际会议
2006中国国际毛纺织会议暨IWTO羊毛论坛(2006 China International Wool Textile Conference & IWTO Wool Forum)
西安
英文
154-161
2006-11-19（万方平台首次上网日期，不代表论文的发表时间）