Predictive Maintenance in Serial Production Using Deep Learning: Analysis of Multi-Layered Telemetry Data Streams and Maintenance History
Management and Production Engineering Review · 02.07.2026 · 17 (2)
DOI: 10.24425/mper.2026.1327 ↗Kurz gefasst
Fachartikel in der Zeitschrift Management and Production Engineering Review, online erschienen am 02.07.2026. Verfasst von Kamil Musiał und 2 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY-NC 4.0).
Abstract (Originalsprache)
This article addresses the problem of predicting machine failures in a company conducting serial, multi-variant production, comprising over one hundred workstations, including welding, spot welding,gluing, and milling. We formulate predictive maintenance (PdM) as a binary early-warning task: attime t, the model predicts whether a failure episode - defined as a computerized maintenance managementsystem (CMMS) corrective intervention causing unplanned downtime of at least 10 minutes -will start within the next H hours (H = 4 h by default; H E {1, 4, 8, 24} h in the sensitivity analysis),using the preceding W hours of telemetry history. The study integrated historical maintenance data,including service requests, Mean Time Between Failures (MTBF), Mean Time To Repair (MTTR),and failure cause codes, with current machine operation telemetry, such as vibration, temperature, current,load, cycle times, and programmable logic controller (PLC) alarms. A complete pipeline was developed,encompassing synchronization of multi-source data streams, feature engineering, and methodsfor addressing class imbalance, such as weighted loss functions and focal loss. The analysis consideredrecurrent neural network architectures - Long Short-Term Memory (LSTM) and Gated RecurrentUnit (GRU) - as well as a one-dimensional convolutional neural network (1D-CNN). Model performancewas assessed using the Area Under the Receiver Operating Characteristic Curve (AUROC),the Area Under the Precision-Recall Curve (AUPRC), and the F1-score, with decision-threshold adjustmentreflecting downtime costs. On the chronologically separated test set, the best-performingLSTM model achieved AUROC = 0.924, AUPRC = 0.847, and F1-score = 0.823, indicating strongpotential for predictive maintenance in serial production.
Zitieren
APA
Musiał, K., Balashov, A., & Burduk, A. (2026). Predictive Maintenance in Serial Production Using Deep Learning: Analysis of Multi-Layered Telemetry Data Streams and Maintenance History. Management and Production Engineering Review, 17(2). https://doi.org/10.24425/mper.2026.1327
Vancouver
Musiał K, Balashov A, Burduk A. Predictive Maintenance in Serial Production Using Deep Learning: Analysis of Multi-Layered Telemetry Data Streams and Maintenance History. Management and Production Engineering Review. 2026;17(2). doi:10.24425/mper.2026.1327
BibTeX
@article{Musia2026,
author = {Musiał, Kamil and Balashov, Artem and Burduk, Anna},
title = {Predictive Maintenance in Serial Production Using Deep Learning: Analysis of Multi-Layered Telemetry Data Streams and Maintenance History},
journal = {Management and Production Engineering Review},
year = {2026},
volume = {17},
number = {2},
doi = {10.24425/mper.2026.1327}
}
EUVIMED