Optimal and Adaptive Control of Stochastic Systems
Project Award Date: 07-01-2012
The control of linear stochastic systems with different cost functionals such as a long run average quadratic cost and an exponential of a quadratic cost is proposed. The systems will be allowed to be partially observed. Other noise processes are also proposed for use in the systems, such as discontinuous processes with correlated increments. Discrete time stochastic systems are proposed for investigation, particularly because they are models for numerical methods. The control of linear stochastic partial differential equations with boundary control and noise and other linear distributed parameter systems that are driven by an arbitrary fractional Brownian motion and that have finite time horizon or infinite horizon time quadratic cost functionals is proposed for investigation. The identification of parameters of a linear stochastic system with an arbitrary fractional Brownian motion is proposed to determine strong consistency and asymptotic normality of some families of estimators. An adaptive control problem for a partially known linear stochastic system driven by an arbitrary fractional Brownian motion is proposed for study extending the proposers results for a scalar linear system.
Primary Sponsor(s): AFOSR