A new algorithm for nonlinear mathematical programming based on fuzzy inference

A new algorithm based on fuzzy inference is proposed for nonlinear mathematical programming. The characteristics of nonlinear programming are that the exact functional relationship between the objective function and the decision variables is unknown; the only information to be utilized about it is described in terms of "If-Then" fuzzy inference rules. Thus, a neuron-fuzzy learning algorithm is adopted to train and learn these fuzzy inference rules so as to determine an approximating function for the objective function. Finally, the proposed theoretical algorithm is applied to a nonlinear programming example and the simulation results are satisfactory.