基于时间序列的纱线直径预测方法

Yarn Diameter Prediction Method Based on Time Series

为了更加精确地针对纱线直径进行预测,更好控制纱线质量,介绍基于使用线阵CCD实时测量纱线直径,运用自回归移动平均模型(ARMA)对纱线直径进行预测即时间序列预测方法的概念、基本思想和预测步骤;对比分析了基于使用自回归模型、移动平均预测模型进行预测的结果。实验结果表明:纱线直径时间序列模型(ARMA模型)预测精度较高,纱线直径时间序列样本存在相关性和平稳性,ARMA时间序列模型在相同样本数量下的精度高于AR模型和MA模型;时间序列预测模型只需参考纱线测量的历史数据,建模简单、预测快捷。

In order to predict the yarn diameter more accurately and control the yarn quality better,the concept,basic idea and prediction steps of the yarn diameter prediction method based on the real-time measurement of the yarn diameter with linear array CCD and using autoregressive moving average model(ARMA),namely the time series prediction method are introduced.The prediction results based on autoregressive model and moving average model are compared and analyzed.The experimental results show that the yarn diameter time series model(ARMA model)has high prediction accuracy,and the yarn diameter time series samples have correlation and stationarity.The accuracy of ARMA model is higher than that of AR model and MA model under the same number of samples.The time series prediction model only needs to refer to the historical data of yarn measurement,so the modeling is simple and the prediction is quick.