Study of Soft-sensing Method Used to Predict the Overhead Product Purity of the Styrene Distillation Column

Abstract

In the paper, a new adaptive soft-sensor model has been used to predict the overhead product purity of the styrene distillation column. Combined with process characteristics of the tower, an adaptive correction method with three deviations is proposed, which based on the recursive PLS. The predictive ability of this soft-sensing method gives accurate predictions of both measured and unmeasured quantities, even in the presence of large unmeasured disturbances. The application results show that this model has a high accurate property. The proposed method also provides a case study to aid in applications to other industrial problems.