用BP神经网络预测精纺毛织物后整理质量

Quality prediction and processing control of the worsted finishing based on BP neural network

通过确定合理的输入层单元数、隐层节点数 ,以 3层BP神经网络建立了原料、纱线和后整理工艺参数与织物质量间的关系 ,对实际加工工艺进行预测 ,获得较满意的效果 ,证明该方法的实用性 ;同时建立了织物质量与工艺参数间的关系 ,并对重要工艺参数进行反演预测 。

A three-layer BP neural network system was established to characterize the relationship between the quality of worsted fabrics and the parameters of wool, yarns and finishing processing through the logical determination of the number of the nodes in the input and hidden layers. A satisfied result has been obtained in the forecast to the parameters of the finishing processing. The validity and significance of the method and its relative models have also been verified by means of the prediction to the industrial data. At the same time, the relation between the quality of fabric and processing parameters is also established to forecast the important parameters, providing instruction for practical processing control.