An innovative strategy for optimal nonlinear feedback control of linear or
nonlinear stochastic dynamic systems is proposed based on the stochastic av
eraging method for quasi-Hamiltonian systems and stochastic dynamic program
ming principle. Feedback control forces of a system are divided into conser
vative parts and dissipative parts. The conservative parts are so selected
that the energy distribution in the controlled system is as requested as po
ssible. Then the response of the system with known conservative control for
ces is reduced to a controlled diffusion process by using the stochastic av
eraging method. The dissipative parts of control forces are obtained from s
olving the stochastic dynamic programming equation.