基于支持向量机算法的造纸过程磨后纤维形态软测量模型

Support Vector Machine Algorithm Based Soft Measurement Model for Post-refiningFiber Morphology of Papermaking Process

中文关键词:  磨浆  纤维形态  软测量  SVM算法

Key Words:pulping process  fiber morphology  soft sensing technology  SVM algorithm

中文摘要:

      本研究提出了一种基于支持向量机（SVM）算法的造纸过程磨后浆料纤维形态软测量模型，该模型利用原始浆板纤维形态参数和磨浆参数作为输入，用于在线软测量生产过程中的磨后浆料纤维形态。结果表明，采用SVM算法进行建模时，7种磨后浆料纤维形态软测量模型的平均相对误差在2.87%~5.61%之间，均优于采用偏最小二乘回归（PLS）算法的建模效果（平均相对误差3.09%~6.60%），模型精度良好，满足生产中纤维形态实时检测对误差的要求。

Abstract:

      In this study, a soft measurement model of post-refining fiber morphology in a papermaking process based on support vector machine algorithm (SVM) was proposed. The model used the parameters of original pulp sheet and refining as input for online soft measurement of post-refining fiber morphology. The results showed that when SVM was used for modeling, the average relative error of the seven kinds of soft measurement models of post-refining fiber morphology was between 2.87% and 5.61%, which was better than the modeling based on PLS algorithm (the average relative error was between 3.09% and 6.60%) , and the model precision was good, which met the error requirements of real-time fiber morphology measurement in production.