| Popis: |
Case studies are reported which show that ideas from stochastic control theory can be exploited to design useful practical controllers. The ideas indicate that the way to design a feedback controller is to implement two separate functions of estimation and control, each matched to some simple mathematical model capturing essential features of the controlled object. Good design does not usually require that either the estimator or the control law be exactly optimal, not even for the simplified model. Where the controlled object is essentially nonlinear, as in many real applications, this suboptimal stochastic design may give better performance than is achievable with classical, black-box linear controllers or their adaptive derivatives. |
