An efficient RNN based algorithm for solving fuzzy nonlinear constrained programming problems with numerical experiments

Abstract

In this study, the solution of the fuzzy nonlinear optimization problems is achieved by a recurrent neural network model. Since there are a few researches for solving fuzzy optimization problems by neural networks, we introduce a new model with reduced complexity to solve the problem. By reformulating the original program to an interval problem and then a weighting problem, the Karush–Kuhn–Tucker optimality conditions are stated. Moreover, we employ the optimality conditions into a neural network as a basic tool to solve the problem. Besides, the global convergence and the Lyapunov stability analysis of the system are debated in this study. Finally, different numerical examples allow to validate our algorithm with the proposed neural network compared to some other alternative networks.