Application Of Principal Component Analysis And Artificial Neural Networks To Rubber Formulation Optimization

Because the complex of industry processing, the method of traditional analysis can not fulfill the demands. Lots of multi-variable, nonlinear redundancy Data is boring the traditional analysis but they are still useful as the recordation of working processing and the quality of production. It is very difficult to analysis all the data with the traditional way.The redundancy information contained in multi-variable data is the origin of trouble to traditional analysis. The Principal Component Analysis (PCA), the most popular dimensionality reduces way people using now, can solve this problem efficiently. But PCA can not deal the problem about nonlinear.As an effective model method, the Artificial Neural Networks (ANN) which doesn't need the exactly modeling had been applied in production extensively. It also has some advantages like excellent learning ability and simulating nonlinear function freely. The problem was that excessive input varies and complex structure often made the convergence accuracy descend and the modeling effect worse.This article studied the advantages of PCA and ANN and designed a new modeling which is artificial neural Networks based on principal component analysis. This modeling can solve the multi-variable, high coupling and redundancy problem among data according the PCA and stimulated the relationship nonlinear mapping among the data.