Elevator Traffic Pattern Recognition Based on Density Peak Clustering

Abstract

Aiming at the shortcomings of traditional methods, this paper proposes an elevator traffic pattern recognition method based on density peak clustering algorithm. This method uses the cluster analysis of the passenger flow data of the previous week to obtain the cluster center coordinates of the corresponding traffic patterns. For real-time changes in elevator traffic data, using 5-minute passenger flow data, the cluster centers are selected based on the highest density and farthest distance from the higher density points, thereby identifying the current traffic pattern. Experiments show that the method can effectively recognize the elevator traffic pattern, is easy to implement, has fast calculation speed, and has a stable clustering effect, and can meet the real-time requirements of the group control system.