Optimization and Simplification of System Models Characterizing an R&d Process

Recent developments in modern control theory and high-speed computation techniques have enabled extensive treatment of complex processes. A method is presented for building a model of an R& D process utilizing the state space approach. The specific model described in this paper contains 21 state variables and a control vector of 6 components. Its formulation leads to the Mayer's problem with inequality constraints imposed on the control vector. An algorithm based on the adjoint system technique is used for simultaneous optimization and simplification. The computation for the preceding model lasts 2.5 min and is completed within 5 iterations. It results in an improvement factor of three for the chosen index of performance. Simplification of the model reduces the 21 state variables and 6 control components into 1 state variable and 1 control component.