Citation: Tnani, M.-A.; Subarnaduti,
P.; Diepold, K. Efficient Feature
Learning Approach for Raw
Industrial Vibration Data Using
Two-Stage Learning Framework.
Sensors 2022, 22, 4813. https://
doi.org/10.3390/s22134813
Academic Editors: Hamed Badihi,
Tao Chen and Ningyun Lu
Received: 30 May 2022
Accepted: 23 June 2022
Published: 25 June 2022
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2022 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
Article
Efficient Feature Learning Approach for Raw Industrial
Vibration Data Using Two-Stage Learning Framework
Mohamed-Ali Tnani
1,2,
* , Paul Subarnaduti
2
and Klaus Diepold
2
1
Department of Factory of the Future, Bosch Rexroth AG, Lise-Meitner-Str. 4, 89081 Ulm, Germany
2
Department of Electrical and Computer Engineering, Technical University of Munich, Arcisstr. 21,
80333 Munich, Germany; ge73gid@mytum.de (P.S.); kldi@tum.de (K.D.)
* Correspondence: mohamed-ali.tnani@boschrexroth.de; Tel.: +49-899-2139-651
Abstract:
In the last decades, data-driven methods have gained great popularity in the industry,
supported by state-of-the-art advancements in machine learning. These methods require a large
quantity of labeled data, which is difficult to obtain and mostly costly and challenging. To address
these challenges, researchers have turned their attention to unsupervised and few-shot learning
methods, which produced encouraging results, particularly in the areas of computer vision and
natural language processing. With the lack of pretrained models, time series feature learning is still
considered as an open area of research. This paper presents an efficient two-stage feature learning
approach for anomaly detection in machine processes, based on a prototype few-shot learning
technique that requires a limited number of labeled samples. The work is evaluated on a real-world
scenario using the publicly available CNC Machining dataset. The proposed method outperforms
the conventional prototypical network and the feature analysis shows a high generalization ability
achieving an F1-score of 90.3%. The comparison with handcrafted features proves the robustness of
the deep features and their invariance to data shifts across machines and time periods, which makes
it a reliable method for sensory industrial applications.
Keywords:
feature learning; CNC machining; machine monitoring; machine learning; few-shot
learning; vibration data; two-stage learning
1. Introduction
The latest advances in technology coupled with an aim to realize smart intelligent
systems have contributed to a rapid move towards the next industrial revolution. Unlike the
third industrial revolution powered by electronics and information technology, digitization
and automation have been the front runners to revolutionize industry to its fourth chapter.
The fourth industrial revolution has proved to be a boon to the traditional machining
processes as it brings some key advantages such as improvement in the production and
quality, cost reduction, and monitoring of machining processes in real time. As a result,
condition monitoring and process condition monitoring systems are integral parts of
intelligent manufacturing that support the quality inspection. Such highly automated
systems rather support the flow of huge volumes of data that can be analyzed in real time
without interrupting any machining workflow [1].
Enabled by the significant advancements in industrial Internet of Things (IIoT), the pro-
cess involved in collecting and monitoring data from industrial environment is made more
convenient. The initial step usually involves the acquisition of different types of signals
such as vibration, cutting force, and a few others that can determine the health of machining
parts and tool processes. This work largely focuses on the vibration-based signals as it
provides critical information about the machining health. However, the vibration signals
collected from the sensors are largely affected by several environmental factors and are
commonly characterized by their nonlinearity, nonstationarity and noisiness. This brings
us to the next steps of monitoring systems that are filtering the collected signals [
2
]. Feature
Sensors 2022, 22, 4813. https://doi.org/10.3390/s22134813 https://www.mdpi.com/journal/sensors
