Design of a simulator for elevator supevisory group controller using ordinal structure fuzzy reasoning with context adaptation

Abstract

An elevator group supervisory controller is a control system that manages systematically two or more elevators in order to serve passengers as required. The elevator cars are assigned accordingly in response to hall calls, so as to optimize waiting time, riding time, power consumption, passengers’ comfort, etc. In order to design a controller that can solve multiple objectives, fuzzy logic would be a good option. However, since in this particular problem, more than three fuzzy inputs have to be considered, complications might arise in forming rule base and fuzzy rule extraction from experts. To overcome this problem, ordinal structured fuzzy logic is to be used where the rules are described in one dimensional space regardless of the number of inputs. In this project, the simplicity of ordinal structured fuzzy logic in making crucial supervisory control decisions is demonstrated. In addition, in order to further improve the performance, a new approach of ordinal structured fuzzy logic with context adaptation is introduced to implement an elevator group supervisory controller for a building with 15 floors and 4 elevator cars. Simulations comparing ordinal structured fuzzy logic algorithm with and without context adaptation, show that the former performs better. An additional improvement is made possible by applying genetic algorithms to tune the weights attached to each of the fuzzy rule.