Research On Quality-related Statistical Monitoring Method For A Class Of Nonlinear Process

Quality-related process monitoring plays an important role in industrial production.The partial least squares(PLS)method can be used to process monitoring by establishing the relationship between process variables and quality variables,which is widely used in the quality monitoring of process control systems.However,in some nonlinear systems,its monitoring capability is poor.Therefore,considering the nonlinear structural features of modeling data is more practical in the process monitoring of practical industrial systems.In response to this problem,based on the expansion of the nonlinear function series,we use the local structure retention feature of local linear embedding(LLE)and the robustness to outliers of L,norm,this paper propose the following two quality-related statistical monitoring methods:1.Considering the nonlinear system as a combination of linear and nonlinear parts,we combined the advantages of linear dimension reduction and local structure retention of PLS method and local linear embedding method(LLE/LPP),a quality-related statistical process monitoring method based on global plus local projection to latent structures(GPLPLS)is constructed.2.Considering that the nonlinear part can also be regarded as the uncertain part of the system,from this point of view,a dual robustness projection latent structure quality-related statistical process monitoring based on L1norm(L1-PLS)is proposed.In order to further improve the robustness of the model,this paper uses the L,norm to obtain the sparse solution and the regression model.The simulation results show that the models constructed by these two methods have good predictive ability,and the corresponding monitoring methods have more realistic fault diagnosis capabilities.