Predictive maintenance of baggage handling conveyors using IoT. (March 2023)

• Record Type: Journal Article
• Authors: Gupta, Vishal; Mitra, Rony; Koenig, Frank; Kumar, Maneesh; Tiwari, Manoj Kumar
• Is Part Of: Computers & industrial engineering. Volume 177(2023)
• Publication Date: 2023-03
• Subjects: Condition monitoring -- Conveyors -- Baggage handling -- Predictive maintenance -- Machine learning; Engineering -- Data processing -- Periodicals; Industrial engineering -- Periodicals; 620.00285
• DOI: 10.1016/j.cie.2023.109033
• Languages: English
• ISSNs: 0360-8352

Abstract

Abstract: This article discusses issues related to the maintenance of airports' baggage handling systems and assesses the feasibility of using predictive maintenance instead of periodic maintenance. The unique issues related to baggage handling systems are discussed — namely random noise captured by the IoT sensors due to the movement of the luggage and complex interconnected components that constitute the conveyors. The paper presents a scalable and economical maintenance 4.0 solution for such a system using data from sensors installed (on a live system in absence of historical data). Differentiating between anomaly detection and outlier detection the paper presents an algorithm that can be used to remove idle and noisy data from the datasets. Using integrated machine learning approaches, it tries to detect and diagnose incumbent defects in the early stage to avoid breakdowns. The paper proposes an automated machine-learning pipeline by processing unstructured industrial data. The performance of various machine learning algorithms on the collected data is compared. Finally, the paper discusses avenues for future research. Highlights: Presented a scalable and economical maintenance 4.0 solution for conveyors used in baggage handling. Introduced an algorithm to remove non-white noisy data from the datasets. Detected and diagnosed incumbent defects in the early stage to avoid breakdowns by using machine learning approaches. Proposed an automated machine learning pipeline by processing unstructured industrial data.