Buch, Englisch, 170 Seiten, Format (B × H): 155 mm x 235 mm
From Unsupervised Anomaly Detection to Few-Shot Fault Diagnosis
Buch, Englisch, 170 Seiten, Format (B × H): 155 mm x 235 mm
Reihe: Engineering Applications of Computational Methods
ISBN: 978-981-9238-25-5
Verlag: Springer
This book presents data-driven methods for unsupervised anomaly detection and few-shot fault diagnosis in complex industrial processes. It is intended for graduate students, academic researchers, and practicing engineers in industrial engineering, automation, and intelligent manufacturing. Complex industrial processes often exhibit strong multivariable coupling, nonlinear dynamics, and long-term temporal dependencies. These characteristics make traditional model-based and rule-based monitoring approaches difficult to apply, particularly when accurate physical models are unavailable and labeled fault data are limited. To address these challenges, the book focuses on two closely related topics: multivariate time-series modeling for unsupervised anomaly detection and meta-learning for few-shot fault diagnosis. The proposed methods are developed for industrial monitoring scenarios and aim to support reliable anomaly detection and intelligent fault diagnosis.
Zielgruppe
Research
Autoren/Hrsg.
Fachgebiete
- Technische Wissenschaften Technik Allgemein Mathematik für Ingenieure
- Technische Wissenschaften Technik Allgemein Technische Instrumentierung
- Mathematik | Informatik EDV | Informatik Informatik Künstliche Intelligenz Maschinelles Lernen
- Technische Wissenschaften Maschinenbau | Werkstoffkunde Produktionstechnik Fertigungstechnik
- Mathematik | Informatik Mathematik Numerik und Wissenschaftliches Rechnen Angewandte Mathematik, Mathematische Modelle
Weitere Infos & Material
Introduction.- Anomaly Detection Based on Improved Diffusion Convolutional Gated Recurrent Unit Network.- Anomaly Detection Based on Spatio-Temporal Modeling with Temporal and Graph Convolution.
