Elevator Maintenance Site Selection Optimization via Fine-Tuned K-Means

Abstract

Elevator failure is a common problem scenario in real life, and arranging reasonable maintenance site locations is crucial to reduce the time to repair elevators. However, reasonable site selection for large-scale datasets remains a challenge in the present time. Considering the nature of clustering models that may naturally fit the requirements of the siting problem, we propose a siting method called Fine-tune K-means which called FTK-means. This approach is based on the K-means clustering method and fine-tunes the initial clustering results. It is worth mentioning that we take into account the fault properties of elevators and use the concept of fault center of mass when finding the mean value. Experiments prove that our method achieves better site selection on real elevator datasets, and in addition achieves site load fairness.