1. Introduction
Elevator systems are the most important transportation method for high-rise buildings. However, in the conventional elevator, only one elevator car (or cage) suspended by ropes occupies the whole elevator shaft, and therefore, the taller the building is, the lower the performance of an elevator becomes. With the background of progress in linear motor technology and increasing needs for high performance transportation systems for large scale buildings, rope-less multi-car elevators (MCE) that have several cars driven by linear motors in a single elevator shaft attract attention as a novel transportation system[1], [6], [7]. However, as for control of MCE, applicability of the knowledge of existing elevator systems to MCE is quite limited, and design methods for controller is needed. Simulation-based optimization of the control strategy is one of candidates for design methods. Sudo et al. have demonstrated that a genetic algorithm can find good control strategies through simulation-based optimization for MCE systems[7]. However, it takes very long computation time because
Evaluation of a control strategy is obtained through complex discrete event simulation of the MCE system, and
Multiple simulation runs are needed so as to reduce random fluctuation involved in an obtained evaluation due to simulation using random numbers.