Process Identification using Nonlinear Programming Techniques

A nonlinear programming approach is developed to estimate process variables and parameters in nonlinear dynamic systems. This approach is used for identification and nonlinear predictive control of a jacketed CSTR with deadtime on the measurements. Studies are performed that illustrate the effects of model structure uncertainty, parameter uncertainty, measurement noise and measurement deadtime uncertainty. This estimation-based approach is shown to be superior to nonlinear predictive control without estimation.