基于PLS和SVM的纸张抗张强度建模比较

Comparison of Paper Tensile Strength Prediction Models Based on PLS and SVM Methods

为解决纸张抗张强度预测模型实际相关性差、预测精度低的问题,基于某瓦楞纸厂生产线,通过机理分析筛选出影响抗张强度的生产变量,分别使用偏最小二乘法(PLS)和支持向量机法(SVM)对抗张强度建模,并通过相关性筛选后的简化模型对模型预测精度进行比较.结果表明,简化后的支持向量机模型更适合纸张抗张强度的现场预测,其均方根误差为321N/m,皮尔逊相关系数为0.909,预测速度快且模型精度较高.

In order to solve the problems of poor practicality and low accuracy of the existing paper tensile strength prediction models , two prediction models respectively based on the partial least-squares （ PLS ） and the support vector machine （ SVM） are established for a corrugated paper mill by selecting parameters affecting paper tensile strength through mechanism analysis .Then, the two models are simplified by deleting parameters of low correlation with tensile strength , and the simplified models are compared in terms of prediction accuracy .The results show that the simplified SVM model , whose root mean square error and Pearson correlation coefficient are 321 N/m and 0.909 respectively , is a quick prediction model with a high accuracy , so it is more suitable for the on-line prediction of tensile strength .