The Model Research Of Main Parts In Process Of Tire Manufacturing

With the rapid development of Chinese automobile industry, the market demand for tire quality is higher and higher. The tread is one of the most important semi-manufactured goods in tire manufacturing, because its quality determinates the performance of tire-manufactured goods. For a tread extruding line in Fengshen Tire Group, HeNan. The paper used the recently ampled data from database of the Access of the tread extruding line to make the research of building model of main part of the tread extruding line procession , based on the Least Squared Support vector machine. It could be made as the basic of suring optimization of control system designing and prediction.The paper analyzed the prime data from Access database and built up a proper model to describ the relation between the inputs consist of 6 test attributes and 2 outputs contain the instantaneous weight and the tire width. The key part of tread extruding line is a complex,nonlinear and dynamic industrial system, and we hadn't known clearly how they work.,so it is usually hard to analysis the mechanism of processes, moreover it will take us unbearable staff cost and financial cost.because of that,we used the Least square support vector machine to analyse the system and build up the model,which come from Statistical learning theory (SLT).Because this paper gathers the primary data contains 19 attributes, we needs to does not need temporarily the attribute value carries on the rejection, and adopts the independent modelling method for these characteristic attributes,such as directions for producing products.Considered from the workload angle that this paper is mainly aims at the data of one direction for producing products to conducting the modelling research. After the above processing,data set includes metrical data under many abnormal condition. For instance: Erroneous data, morbid state data, lag data and so on. This article mainly adopts the artificial elimination method. Then uses the upper and lower limit method as well as the 7-points centered moving average method to reduce the opportunity error and the random error in the data.In order to simplify the input attribute number, this paper adopt PCA (Principal Components Analysis) method to reduse the dimension. Thus 6 input attributes will reduce to 4 that are able to express the overall information of the original input variables. Finally use Least square support vector machine (LS-SVM) method to build up the system model , to reflect the relation between 4 inputs and 2 outputs contain the instantaneous weight and the tire width.Support vector machines (SVM) toolbox in Matlab7.1 environment is introduced,and an extensively overview of the entire collection of toolbox function used to support vector regression is given and we illustrate how to solve the regression problem of modeling.At last we analysed the result figures of simulation.