Optimal Controllers for Stochastic Jump Processes.
Abstract
The report discusses the optimal feedback Bayes control, with memory, of a dynamic system governed by the statistics of any stochastic regular jump process. The latter is characterized in terms of the associated regular point process and the embedded state sequence. An optimality theorem, which gives a sufficient condition for an admissible control function to be optimal, is derived. A separation property, concerning the control and filtering operations, is established for the Bayes controller. The results are applied to derive optimal controllers for linear systems with random failures and renewals, assuming the latter to follow an alternating renewal point process model, or to be governed by a binary Markov process with unknown transition intensities. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 01, 1972
- Accession Number
- AD0756224
Entities
People
- Izhak Rubin
Organizations
- University of California, Los Angeles