Abstract:

This paper analyzes and sums up the characteristic attributes of elevator group-control system with destinationregistration according to the new flow distribution modes of the elevator system , and proposes a new elevatorgroup-control policy based on multi-objective fuzzy neural network. The proposed policy regards the characteristicattributes of the elevator system as the inputs to establish a multi-objective fuzzy neural network , which is thentrained and used as the elevator group-control calculator. This novel elevator group-control calculator can be used todispatch elevators. Simulated results show that the proposed policy is of excellent performance. Moreover , as comparedwith the traditional elevator group-control policy and some other elevator group-control policies with destinationregistration , the proposed policy behaves better in average waiting and riding time , long-waiting ratio , long-ridingratio and power consumption , especially in down peak traffic pattem.

Key words: destination registration, multiple objective, fuzzy neural network, elevator group-control system