A Blind Condition-Based Maintenance Framework for Real-Time Fault Detection and Degradation Modeling of the LINK APM Gearbox

Automated people movers (APMs) are a critical piece of infrastructure in many airport operations, providing passengers, and airport personnel with an efficient means of travel between terminals, connections, or parking facilities. Reliability and sustained uptime of these APMs are paramount for efficient operation. Traditional maintenance strategies which employ routine maintenance can result in unnecessary, unwanted system downtime, as well as wasted labour and material resources. Preventative maintenance, in the form of condition-based maintenance (CBM), is an alternative maintenance strategy that circumvents the pitfalls of traditional strategies. CBM works by continuously monitoring the APM system for any signs of an incipient fault, and promptly notifying maintenance personnel when a fault is detected. In other words, maintenance is only performed on an as-needed basis, maximizing both system uptime, and distribution of resources. However, current CBM approaches are not without their own drawbacks. These approaches often focus on monitoring the system as a whole, rather than monitoring individual components. Furthermore, the accuracy of these approaches is heavily dependent on the availability of historical data. In this paper, a robust, blind CBM framework for monitoring of the LINK APM gearbox is presented. The proposed framework utilizes vibration measurements coupled with a novel signal pre-processing algorithm to detect incipient faults and build degradation models of the monitored system. Contrary to the current body of CBM approaches, the proposed approach is capable of monitoring and modeling the degradation of different families of components (i.e., gears and bearings) separately, allowing for increased detection and life-cycle prediction accuracy. Additionally, the proposed framework requires no historical data or prior knowledge of the system: detection accuracy and confidence in the degradation model parameters gradually increases as new sensor data becomes available. The underlying pre-processing algorithm has been validated using data collected from a number of real industrial systems and was shown to perform well under a wide variety of operating conditions. The blind, computationally efficient nature of the pre-processing algorithm, coupled with its minimal hardware requirements, allows the proposed CBM framework to be easily implementable on any APM system or rotating machinery asset.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.