Fast Computational Method for a Class of Nonlinear Bilevel Programming Problems Using the Hybrid Genetic Algorithm

In this paper, a fast computational method for a class of nonlinear bilevel programming problems is proposed. In these problems, the lower-level problem can be decomposed into some paratactic and independent sub-problems. First, by Karush-Kuhn-Tucker optimality, the stationary-points of these sub-problems corresponding to the upper-level variables can be determined. As a result, this kind of nonlinear bilevel programming is transformed into a single level optimization problem. The hybrid genetic algorithm is then adopted to solve this single optimization problem. Simulation results on 18 benchmark problems show that the proposed method is able to solve effectively the bilevel programming problems such that their global optima can be found, with high convergent speed and less computational cost compared to other existing algorithms