Elevator Group Control with Artificial Intelligence

In this report a novel control that optimizes passenger service in an elevator group is described. Fuzzy logic and artificial intelligence are applied in the control when allocating landing calls to the elevators. Fuzzy logic is used to recognize the traffic pattern and the traffic peaks from statistical forecasts. In order to form the statistical forecasts, the passenger traffic flow in the building is measured. In the statistics the passenger traffic flow is learned day by day, and the control adapts to the prevailing traffic situation. The validity of the forecast data is confirmed before applying it in the control. The measured passenger arrival rate at each floor is used in optimizing the passenger waiting times and the ride times inside a car. The methods developed in this work are utilized in real group controls. With these controls average waiting times and waiting time distribution floor by floor are decreased compared with conventional controls that optimize landing call times. An example of an office building where an old electronic control was modernized with the TMS9000 control shows a considerable improvement in the resulting landing call times.