Computational Nonlinear Stochastic Control
Academic Article
Overview
Identity
Additional Document Info
Other
View All
Overview
abstract
A computational methodology for solving the nonlinear stochastic optimal control based on the policy iteration algorithm and a finite dimensional approximation of the controlled Fokker-Planck-Kolomogorov/diffusion operator has been recorded. The method of successive approximations/policy iteration has widely been used to solve deterministic optimal control problems. An approximate solutions of the Fokker-Planck-Kolmogorov equation (FPE) are considered and leveraged for the design of controllers for nonlinear stochastic dynamic systems. The policy iteration algorithm can be used either offline or online because the sequence of controllers obtained are monotonically improving with respect to the defined cost function. The computational methodology was on a number of test cases where it revealed to have satisfactory performance.
