Hybridized Genetic ALgorithm with Neural Network Technique for Solving Non-linear Mixed Integer Programming Problems

摘要

In this paper, we discuss non-linear mixed integer programming (NMIP) models which should be simultaneously determined continuous and discrete decision variables. This problem is more difficult than the NIP problem while more actually representing the real world. Recently, several researchers have obtained acceptable and satisfactory results by using genetic algorithms for NMIP problems. For large size problems, however, genetic algorithm (GA) has a lot enumeration of feasible solutions due to broad continuous search space, i.e., it require too large computational time and computer memroy. To overcome this kind of difficulties. we propose a hybridized GA combined neural network (NN) technique suitable for approximating continuous optimal solutions. Numerical exper ments and comparison with the previous works demonstrate the efficiency of our proposed method.