Nonlinear Optimal Control of Two-Level Large-Scale Systems; Part I - Interaction Prediction Principle

Abstract

In this paper, which consists of two parts, a new two-level computational algorithm is used for nonlinear optimal control of large-scale systems. The two-level optimizer uses a new coordination strategy which is based on the gradient of interaction errors instead of the gradient of overall performance function. The advantages of the new method can be categorized into two parts: First, the new formulation is applicable to a large class of problems whilst the classical model coordination method is not. Second, it extremely reduces the number of iterations required for obtaining the overall optimal solution. Although the computational burden of the algorithm in the first level increases slightly, it is shown that there is a great saving in computation time of the overall problem. To demonstrate the significance of the theoretical developments and the computational requirements for new algorithm, a numerical example is presented.