国毛毛条加工性能和质量预测

Prediction of top processing performance and quality on domestic wool

通过对国产细羊毛原毛特征和毛条加工性能的研究,运用多元回归统计和人工神经网络两种方法建立了国毛毛条加工性能和质量的预测模型.实践表明:两种模型对毛条的豪特长度、长度变异、短纤维含量和精梳落毛率均能进行较为准确的预测,其中人工神经网络预测模型预测效果优于统计回归模型.

Based on studies of characteristics of domestic greasy wool and top-processing performance two approaches with statistics and artificial neural network procedures are applied to establish top processing performance and quality prediction formulae which the top Hauteur, coefficient of variation of Hauteur, short fibre content and combing N_(oil) can be predicted. Application experience has been shown that both methods have good prediction performance in which the ANN method has better prediction performance in comparison with multiple regression method.