Robust nonlinear PLS based on neural networks and application to composition estimator for high-purity distillation columns

The accurate and reliable on-line estimation of product quality is an essen tial task for successful process operation and control. This paper proposes a new estimation method that extends the conventional linear PLS (Partial Least Squares) regression method to a nonlinear framework in a more robust: manner. To handle the nonlinearities, nonlinear PLS based on linear PLS an d neural network has been employed. To improve the robustness of the nonlin ear PLS, the autoassociative neural network has been integrated with nonlin ear PLS. The integration allows us to handle the nonlinear correlation as w ell as nonlinear functional relationship with fewer components in a more ro bust manner. The application results have shown that: the proposed Robust N onlinear PLS (RNPLS) performs better than previous linear and nonlinear reg ression methods such as PLS, NNPLS, even for the nonlinearities due to oper ating condition changes, limited observations, and measurement noise.