色织生产调度中基于ANFIS的整经轴数智能预测

ANFIS-based Method for the Prediction of Trim-Beam Number in Colored-textile Manufacturing Process Scheduling Problem

分析了整经轴数预测在实际色织生产过程调度中所起的重要作用。由于与整经轴数相关的工艺属性较多且关系复杂,一般的神经网络法难以确定工艺属性与整经轴数之间关系或应用效果不佳,同时神经网络是一个“黑箱”,从中无法对影响整经轴数的因素进行分析,因此提出了基于ANFIS的整经轴数智能预测方法。在该方法中,采用ANFIS作为整经轴数的预测技术,并针对ANFIS输入变量维数较大的问题,提出了一种基于模糊C均值聚类算法的ANFIS结构辨识方法。将该预测方法用于实际的色织生产过程调度的整经轴数预测中,实验结果表明该预测方法是有效的。

Prediction problem of trim-beam number in colored-textile manufacturing process scheduling is discussed. Because there are many properties related with trim-beam number and the neural networks method is only a ‘black box＇, it is difficult to identify the relationship between properties and trim-beam number by using neural networks. An ANFIS-based method for the prediction of trim-beam number is proposed. In this method, practical manufacturing process data is used to train the ANFIS, and this trained ANFIS is then used to predict the trim-beam number. Meanwhile, aiming at the modeling difficuhy of ANFIS with many inputs, the structure identification technique of ANFIS based on bintree linear partition is improved with fuzzy c-means algorithm. The prediction method is applied to the practical colored-textile mantrfacturing process scheduling problem, and the comparing results show that this ANFIS-based prediction method is effective.