Optimal Controller Placements in Large Scale Linear Systems

The problem of controller placements in large scale linear systems is considered in this paper. Specifically, given a large scale linear system, it is desired to find the optimal number of linear controllers, the optimal locations in the system to place these controllers and the optimal feedback gain of each contoller so that an objective function is optimized, and also so that other desirable properties of the controlled system, such as eigenstructure assignment, occur. A formulation of the controller placement problem is presented. A general solution algorithm based on simulated annealing is developed for the controller placement problem. The solution algorithm can arrive at the global optimal solution. The solution algorithm has been implemented into a software package and tested on a 132-dimension linear system with very promising results.