Simulated annealing based global optimization of the general bilevel nonlinear programming problems
Zhou Li-ni · Systems engineering and electronics · 2004
A global approach based on simulated annealing (SA) for solving the general bilevel nonlinear programming problems is proposed. The higher-level problem is solved by simulated annealing and the accurate penalty function is employed to deal with its constraints, which can ensure the global solutions are got rapidly and steadily. In order to improve the algorithm efficiency, some improvement measures are applied to the standard SA. The varied-tolerance simplex method is used to solve the lower-level nonlinear programming problems. The combinational approach proposed has a clear idea and can be easily programmed. Its high reliability of global convergence and high convergence rate have been testified by the numerical computations, and it is an effective approach in solving the general bilevel nonlinear programming problems.