In this paper, a method of designing control inputs for stochastic nonlinear processes under state-feedback is proposed. The objective is to determine a control input that minimizes the expected value of the integral of error between the set-point and the states. Since the states may not be measured, they are estimated using a particle filtering algorithm. The optimal control design is then reformulated as a parameter estimation problem using control vector parameterization where the inputs are considered as a nonlinear function of the error between the state estimates and the set-point. The parameters are then computed through a homotopy based optimization method. The control performance resulting from proposed homotopy based optimization method is compared with that of direct optimization and an existing nonlinear control method on a Solid Oxide Fuel Cell (SOFC) stack model. © 2021, The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature.