Very efficient hybrid simulation methods for complex problems

Simulation is a powerful tool to explore the real-world systems; unfortunately, two shortcomings frequently accompany simulations. First, simulation generally requires long computation (simulation) time. Second, the quality of found solutions are unknown. In other words, it is hard to accurately estimate how long a simulation should be run to find the optimal solution to a problem, and it is equally difficult to assess how good these solutions are even after a long simulation time. The purpose of this dissertation is to study hybrid simulation methods, i.e., to combine other methods with simulation to reduce simulation time and use ordinal optimization to evaluate the quality of the best solution found during a simulation. These hybrid simulation methods include simple heuristic methods, evolutionary computation methods, and distributed computing methods for different types of complex problems. Examples in Chapter 3 to Chapter 5 will confirm the efficiency of these hybrid simulation methods