基于最小二乘支持向量机的水松纸透气度检测研究

发布时间：2018-03-26 08:48

本文选题：水松纸　切入点：透气度　出处：《昆明理工大学》2016年硕士论文

【摘要】：打孔水松纸透气度的大小严重制约卷烟中焦油的含量,而焦油又是香烟对人体造成伤害的主要成分,随着健康问题日益成为人们关注的焦点,水松纸透气度检测也正在逐步成为烟草行业的技术重点,关系着是否能在烟草行业立足,是人们健康问题的根本保障。因此,水松纸透气度检测分析是当前一直被关注的研究领域,构建有效的水松纸透气度软测量模型对水松纸透气度实现高效检测具有重大意义。本文根据已设计的基于图像的新型透气度检测设备,结合传统打孔水松纸透气度检测原理和水松纸图像信息,分析了影响打孔水松纸透气度的主要因素为打孔水松纸打孔面积和水松纸灰度值,进而采用单输入(打孔水松纸打孔面积)、双输入(打孔水松纸打孔面积、水松纸灰度值)两种方式对水松纸透气度进行建模并实现检测。由于支持向量机(Support Vector Machine,SVM)具有能够较好地解决小样本、非线性、高维数问题的特性,故本文采用SVM建立软测量模型并对水松纸透气度进行检测。首先,基于对支持向量机(Support Vector Machine,SVM)的研究,通过利用训练误差的平方代替松弛变量,将不等式约束改为等式约束,从而提出基于最小二乘支持向量机(Least Squares Support Vector Machine,LSSVM)的水松纸透气度软测量模型。这可避免求解二次规划问题,提高了检测模型的训练速度。其次,考虑到LSSVM水松纸透气度软测量模型的参数对检测结果精度有着至关重要的影响,为避免模型参数选择的盲目性,提高模型的泛化能力,本文利用粒子群优化算法(Particle Swarm Optimization,PSO)对LSSVM中的参数进行确定,得到基于PSO-LSSVM的水松纸透气度软测量模型。基于实际数据的仿真实验表明,所提的单输入和双输入模型均能都得到更好的检测效果。最后,为了进一步提高单一模型的检测精度和泛化能力,在PSO-LSSVM的基础上,结合集成学习方法,提出了一种改进的AdaBoost-PSO-LSSVM水松纸透气度软测量模型。基于实际数据的仿真测试表明,所提出的基于改进的AdaBoost-PSO-LSSVM比PSO-LSSVM具有更高的检测精度。采用论文所建立软测量模型的新型透气度检测设备已在国内某水松纸厂得到了实际运用,效果良好。
[Abstract]:The degree of air permeability of perforated pine paper seriously restricts the content of tar in cigarettes, and tar is the main component of cigarette harm to human body, with the health issues increasingly becoming the focus of attention. The measurement of air permeability of pine paper is gradually becoming the technical focus of tobacco industry, which is related to whether it can be established in the tobacco industry, and is the fundamental guarantee of people's health problems. Air permeability detection and analysis of water pine paper is a research field that has been paid close attention to at present. It is of great significance to construct an effective soft sensing model for air permeability measurement of water pine paper. Based on the principle of air permeability detection and image information, the paper analyzes that the main factors that affect the air permeability of the paper are the perforating area and the gray value of the paper. Then the single input (perforating water paper perforation area), double input (perforating water paper perforation area, Because the support vector machine (SVM) can solve the problem of small sample, nonlinear and high dimension, it can solve the problem of small sample, nonlinear and high dimension, because the support vector machine (SVM) can solve the problem of small sample, nonlinear and high dimension. In this paper, SVM is used to establish a soft sensing model and to detect the air permeability of water pine paper. Firstly, based on the research of support Vector Machine (SVM), the inequality constraint is changed into an equality constraint by using the square of training error instead of the relaxation variable. Thus, a soft sensing model for air permeability of water pine paper based on least squares support vector machine (Least Squares Support Vector Machine) is proposed, which can avoid solving quadratic programming problem and improve the training speed of the detection model. Considering that the parameters of the soft sensing model for the air permeability of LSSVM water pine paper have an important influence on the accuracy of the test results, in order to avoid the blindness of model parameter selection and improve the generalization ability of the model, In this paper, particle swarm optimization algorithm (PSO) is used to determine the parameters of LSSVM, and a soft sensing model of air permeability of water pine paper based on PSO-LSSVM is obtained. In order to further improve the detection accuracy and generalization ability of the single model, the proposed single input and double input models can both get better detection results. Finally, based on PSO-LSSVM, an integrated learning method is used to improve the detection accuracy and generalization ability of the single model. An improved soft sensing model for air permeability of AdaBoost-PSO-LSSVM water pine paper is proposed. The simulation results based on the actual data show that, The improved AdaBoost-PSO-LSSVM has higher detection accuracy than PSO-LSSVM. A new type of air permeability testing equipment based on soft sensing model has been applied in a domestic pines mill with good results.
【学位授予单位】：昆明理工大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TS77;TP18

【相似文献】