Development of a statistical process control method using independent component analysis and support vector machine

Abstract

Soft sensors are widely used to estimate a process variable which is difficult to measure online. One of the crucial difficulties of soft sensors is that predictive accuracy drops due to changes of state of chemical plants, catalyst performance loss, sensor and process drifting, and so on. In order to cope with this problem, a regression model can be updated with a new sample. However, if the model is updated with an abnormal sample, the predictive ability can deteriorate. We have applied the independent component analysis (ICA) method and support vector machine (SVM) to the soft sensor in order to increase fault detection and diagnosis ability. Then, we have tried to increase the predictive accuracy. By using the ICA and SVM based fault detection and diagnosis model, the objective variable can be predicted, updating the regression model appropriately. We analyzed real industrial data as the application of the proposed method. RMSE of a traditional soft sensor model are 0.511, and that of the proposed soft sensor model are 0.200. The proposed method achieved higher predictive and fault detection accuracy than the traditional one.