Study On Air Permeability Detection Of Tidal Paper Based On Least Squares Support Vector Machine

The porosity of tipping paper is severely restrict the tar content of cigarettes, and cigarette tar is a main component which injured our bodily, With health issues increasingly become the focus of attention, tipping paper porosity detection is gradually becoming the key technology of the tobacco industry’s, it’s related that whether they can gain a foothold between the tobacco industry, it’s also the fundamental guarantee of people’s health. Therefore, the detection and analysis of tipping paper porosity is the field which has been concerned, it has a great significance to build an effective soft sensor model of tipping paper porosity, which can achieve efficient detection in tipping paper porosity.This paper is based on the new tipping paper porosity testing equipment which has been designed based on image, it combined with the detection principle of traditional perforated tipping paper porosity and the image information of tipping paper, analyzed the main factors that affect the porosity of perforated tipping paper are the perforated area and the gray value of perforated tipping paper, further, the tipping paper porosity test model was been established by single input (the perforated area of perforated tipping paper), dual input (the perforated area of perforated tipping paper, the gray value of tipping paper), it can implement test in both two ways. Therefore, this paper established tipping paper porosity soft sensor model using support vector machine (Support Vector Machine, SVM) according to SVM has the properties to solve small sample, nonlinear, high dimension problem.Firstly, Support Vector Machine (Support Vector Machine, SVM)has been studied in this paper, by using the square of training error instead of the slack variable, the inequality constraints will be changed to equality constraints, then this paper put forward a soft sensor model of tipping paper porosity by using LS-SVM (Least Squares Support Vector Machine, LSSVM). This can avoid solving quadratic programming problems and improve the training speed of detection model.Secondly, taking into account that the parameters of tipping paper porosity of LSSVM soft measurement model has a critical influence in the accuracy of test results, in order to avoid the blindness in choosing model parameters, improve the generalization ability of the model, in this paper, PSO (Particle Swarm Optimization, PSO) is used to determine the parameters of LSSVM, then get the soft sensor model of tipping paper porosity based on PSO-LSSVM. Simulation results based on the actual data show that both single input and dual input model which has mentioned could have a better detection results.Finally, in order to further improve the detection accuracy and generalization ability of single model, an improved AdaBoost-PSO-LSSVM soft sensor model of tipping paper porosity has proposed, which is based on PSO-LSSVM, combining integrated learning method. Simulation results based on the actual data show that the AdaBoost-PSO-LSSVM proposed in this paper has higher detection accuracy than PSO-LSSVM.The new porosity testing equipment of soft sensor model in this paper has practical applied in the domestic mills and has a good result.