摘要
采用BP神经网络预测手摘籽棉脱籽后原棉的多个品质指标。以南疆库尔勒地区某轧花厂为试验场,选择籽棉回潮率、轧花速度和喂花量3个关键因素作为输入层自变量,建立手摘籽棉的3个轧花品质指标(UHML、Str、SF)预测模型。试验结果表明:3-3-2-2-3网络拓扑结构的BP神经网络模型能较好表达轧花原棉3个品质指标与关键因素之间的复杂关系,轧花品质指标预测模型的预测输出值与试验目标值的相关系数接近0.984 1,BP神经网络模型可实现籽棉多个轧花品质指标的有效预测。
BP neural network was used to forecast multiple quality indexes of raw cotton after de-seeding.A ginnery in Korla of south Xinjiang was chosen as the test field.3 key factors(moisture regain of unginned cotton,cotton ginning speed and feeding volume)were selected as independent variables of input layer to establish the forecasting model(involving 3 quality indexes UHML,Str,SF).The results show that the BP neural network model(3-3-2-2-3 model)can well express the complex relationship between 3 quality indexes and key factors.The correlation coefficient of predicted output value of quality index forecasting model and test target value is close to 0.9841.BP neural network model can achieve effective prediction of multiple cotton quality indexes.
出处
《现代纺织技术》
2018年第1期49-54,共6页
Advanced Textile Technology
基金
国家质检总局项目(201310107)
新疆生产建设兵团支疆项目(2012AB008)
塔里木大学校长基金自然科学项目(TDZKSS201322)
关键词
BP神经网络
原棉
预测模型
品质指标
多隐含层
BP neural network
raw cotton
forecasting model
quality index
multiple hidden layers
