A decomposition technique for discrete time optimal control problems with an application to water resources management

  • Authors:

  • T. A. Edmunds;J. F. Bard

  • Affiliations:

  • Operations Research Group, Department of Mechanical Engineering, University of Texas at Austin, Austin, TX 78712, U.S.A.;Operations Research Group, Department of Mechanical Engineering, University of Texas at Austin, Austin, TX 78712, U.S.A.

  • Venue:
  • Mathematical and Computer Modelling: An International Journal
  • Year:
  • 1990

Quantified Score

Hi-index 0.98

Visualization

Abstract

The purpose of this paper is to present a new decomposition technique that can be used to solve large-scale optimal control problems. The focus is on models that are either too large to fit into available computer memory, or too unwieldy to be solved with standard nonlinear programming routines. Such models often arise from discrete approximations to continuous formulations. Our approach reduces both computational and storage requirements by creating a series of easy-to-solve subproblems through time axis decomposition. Information obtained from the resulting solutions is used to construct an optimal trajectory for the original problem. The technique is illustrated with an application concerned with the scheduling of a three-reservoir hydroelectric power system. The analysis highlights the computational requirements of the methodology as well as some of its advantages and shortcomings.