Investigating thresholding techniques in a real predictive maintenance scenario

Abstract

We deal with a real predictive maintenance (PdM) scenario in an Industry 4.0 setting. With a help of the Sibyl platform, we can monitor live data from key components of a Philips factory equipment; in this work, we focus on a cold-forming press. Due to the dynamic environment of the operation of this press, unsupervised anomaly detection techniques are used to timely detect the wear, where early anomalies are interpreted as warning signs of a forthcoming failure. Typically such techniques assign an anomaly score, and the problem we face is how to appropriately set a threshold for this score. We introduce and compare four generally applicable thresholding techniques, two of which are dynamic, i.e., they continuously refine the threshold during the episode lifetime. We discuss the properties of these techniques and quantitatively evaluate their behavior in our case study.

AI in renewable energy: A review of predictive maintenance and energy optimization

“…In the context of renewable energy, Deep Learning algorithms can extract valuable insights from various sources, including image data from surveillance cameras or drones (Capra et al, 2020). This allows for the detection of visual anomalies or signs of wear and tear on equipment, enabling preemptive maintenance actions (Giannoulidis et al 2022). Meanwhile, the concept of Digital Twins further amplifies the effectiveness of predictive maintenance.…”

Introduction to the Special Section on AI in Manufacturing

“…Revised contributions of five article were invited for this special section. These five contributed articles span a variety of topics: predictive maintenance [1], fault detection and prediction [2], (acoustic) structural integrity assessment [3], industrial time-series classification [4], and supply chain link prediction [5]. Additionally, a poster was presented on few-shot learning to identify faults and concept drift, and there were two demo's.…”

“…Investigating thresholding techniques in a real predictive maintenance scenario Giannoulidis et al [1] present a study about unsupervised anomaly detection for predictive maintenance, on real streaming data coming from a coldforming press. Anomaly detection methods frequently output a score, e.g., a probability, and then employ a threshold to decide whether the current data window warrants throwing an alarm message.…”