Hybridized neural network and genetic algorithms for solving nonlinear integer programming

Optimization problems such as system reliability design and general assignment problem are generally formulated as a nonlinear integer programming (NIP) problem. Generally, we transform the nonlinear integer programming problem to a linear one in order to solve NIP problems. However linear programming problems transformed from NIP problems become a large-scale problem. In principle, it is desired that we deal with the NIP problems without any transformation. In this paper, we propose a new method where a neural network technique is hybridized with genetic algorithms for solving nonlinear integer programming problems. The effectiveness and efficiency of this approach are shown with numerical simulations from the reliability design problem.