방적

Applications of Neural Networks in Spinning Prediction.

  • 출판일2002.05
  • 저자
  • 서지사항
  • 등록일 2016.11.02
  • 조회수 224
A neural network simulation of yarn spinning predicts yarn quality and spinning performance via backpropagation and radial basis function networks. Inputs include count, diameter, hauteur, bundle strength, spinning draft, traveler number, and twist. Outputs include yarn evenness, thin places, tenacity, elongation, and ends down. Comparisons between the predicted values and experimental data revealed that errors in predicting coefficient of variation were less than 1 percent. Errors in predictions of thin places and tenacity were less than 5 percent. The error rate for prediction of ends down was higher but acceptable considering the number of variables and the limited number of groups of data in training. 5 refs.
이전글 다음글 목록