Optimal instantaneous output-feedback controllers for linear stochastic systems

Abstract

The problem of determining a linear feedback control that depends only on the instantaneous value of the system output is solved for a class of linear stochastic; systems with quadratic cost criterion. For the finite time (steady-state) version it is shown that a two-point boundary-value problem (two non-linear algebraic equations) must be solved to implement the optimal control. The iterative methods to solve the above equations are briefly reviewed. A number of examples solved on a digital computer are included in the paper.