Approximate solutions of fuzzy optimal control problems using sigmoid-weighted neural networks

Abstract

Optimal control problem is one of the most challenging subjects in control theory. It has numerous applications in science and engineering. In this study, we are motivated to obtain the solution of fuzzy optimal control problems via universal approximation capability of a single-layer feedforward artificial neural network. First, we transform the fuzzy optimal control problems into systems of first-order ordinary differential equations via fuzzy Pontryagin’s minimum principle and fuzzy Hamiltonian function. Then, we solve these systems by using a single-layer feedforward sigmoid-weighted neural network. The numerical examples are presented to determine the simplicity and efficiency of the proposed method.