STATISTICAL CONTROL SYSTEM DESIGN.
Final rept., May 65-May 66,
GRUMMAN AIRCRAFT ENGINEERING CORP BETHPAGE N Y RESEARCH DEPT
Pagination or Media Count:
The report discusses the performance of optimal control systems with random parameters. The plants treated are described by time-invariant ordinary differential equations whose parameters are random variables with known distributions. The control law remains fixed as the nominal optimal control, and it is assumed known. Because the parameters of the differential equations are random variables, the performance is a random variable and can only be described in terms of its statistics. Several methods of evaluating the ensemble statistics are presented and compared. Alternatives better suited to stochastic systems are given to the classical notion of sensitivity. Numerical results are presented for specific examples, and the computer programs used in the study are discussed. The report also includes an analysis of a particular optimal stochastic control problem, to indicate some of the difficulties involved in deriving optimal stochastic control laws. Author
- Operations Research