A computational approach to the combined problem of optimization and parameter identification

  • Authors:

  • Yacov Y. Haimes;David A. Wismer

  • Affiliations:

  • Associate Professor of Engineering, Systems Engineering Division, School of Engineering, Case Western Reserve University, Cleveland, Ohio USA;Operations Research Division, Systems Control Inc., Palo Alto, California USA

  • Venue:
  • Automatica (Journal of IFAC)
  • Year:
  • 1972

Quantified Score

Hi-index 22.15

Visualization

Abstract

Two important areas in systems engineering are process identification and optimization. The optimization problem depends on a mathematical model which in turn is determined by the identification procedure. Each of these problems is computationally difficult, a feature which has led to separate treatment in the literature. This paper attempts to examine the interdependence of these two problems and to develop an iterative computational approach for solving the joint problem which actually arises in practice; namely, the optimization of a plant whose parameters are in fact unknown. An integrated formulation of the two problems based on the theory of multilevel optimization exploits the mathematical similarity of the two problems. Several methods of decomposing the integrated problem into subsystems representing the optimization and identification problems are derived. An example problem is solved and computational results are presented.