SIMULTANEOUS SOLUTION AND OPTIMIZATION STRATEGIES FOR PARAMETER-ESTIMATION OF DIFFERENTIAL-ALGEBRAIC EQUATION SYSTEMS

This paper addresses the development of a new hybrid successive quadratic programming (SQP) method for solving nonlinear parameter estimation problems. Here, we show that the Hessian information for quadratically convergent algorithms can be constructed using only first derivatives for small residual problems. Furthermore, it has been combined with a tailored quasi-Newton update formula to deal with large residuals. Moreover, the properties of reduced spaces have been exploited to deal very efficiently with large-scale problems. These strategies have also been coupled to a large-scale modeling system (GAMS) which allows for the easy construction of complex process models. Here, a dynamic model is used to illustrate one of the possible applications of the method. The model is discretized by orthogonal collocation on finite elements. With this approach, both solution and optimization occur simultaneously, and constraints on the state variables can be imposed directly.