An Internet of Things-Based Lift Predictive Maintenance System

Lift (or elevator) breakdowns cause huge inconveniences to city dwellers and affect more than 80% of Singapore's residents who live in high-rise apartments. Unfortunately, lift maintenance today is either preventive (e.g., follows a fixed cycle based on statistics) or reactive (the faults are fixed after they happen). In this article, we describe a lift-model-agnostic and nation wide predictive maintenance system for lifts in high-rise apartments in Singapore. Based on real sensor and operational data collected over a 14-month period from more than 100 lifts, our prediction model is able to forecast a breakdown better than random guesses despite a highly imbalanced data set.