Artificial Neural Networks in Elevator Dispatching

This paper describes a "real world" application of neural networks that has enhanced the control of elevators in high rise buildings. A range of artificial neural networks (ANNs) - from a simple perceptron to complicated networks with hidden layer - have been developed and installed in advanced technology elevator systems. It has been demonstrated that the ANNs provide improved accuracy in estimating the response time for a given elevator to reach the floor of a waiting passenger. This greater accuracy has led to improvement in group control performance as measured generally by passenger waiting times. This paper defines the elevator dispatching problem and some of the research work that has led to the implementation of a neural network based elevator dispatcher in the real world.