Autoencoder-based Representation Learning from Heterogeneous\n Multivariate Time Series Data of Mechatronic Systems

2021/04/06 by Karl-Philipp Kortmann, Kortmann, Karl-Philipp, Moritz Fehsenfeld +3 · 1 citation
Computer Science · Engineering · #68T05 (Primary) 62H12 #68T07 (Secondary) #Advanced Chemical Sensor Technologies #Anomaly Detection Techniques and Applications #Artificial Intelligence (cs.AI) #FOS: Computer and information sciences #FOS: Electrical engineering #J.2 #Machine Learning (cs.LG) #Signal Processing (eess.SP) #Time Series Analysis and Forecasting #electronic engineering #information engineering

paper · pdf · doi:10.48550/arxiv.2104.02784

openalex publication_date 2021/04/06 · openalex created_date 2022/07/25 · openalex updated_date 2026/07/28

Abstract

Sensor and control data of modern mechatronic systems are often available as\nheterogeneous time series with different sampling rates and value ranges.\nSuitable classification and regression methods from the field of supervised\nmachine learning already exist for predictive tasks, for example in the context\nof condition monitoring, but their performance scales strongly with the number\nof labeled training data. Their provision is often associated with high effort\nin the form of person-hours or additional sensors. In this paper, we present a\nmethod for unsupervised feature extraction using autoencoder networks that\nspecifically addresses the heterogeneous nature of the database and reduces the\namount of labeled training data required compared to existing methods. Three\npublic datasets of mechatronic systems from different application domains are\nused to validate the results.\n

Citations

• MAD-GAN: Multivariate Anomaly Detection for Time Series Data with Generative Adversarial Networks

Cited by

• Tool-to-Tool Matching Analysis Based Difference Score Computation Methods for Semiconductor Manufacturing

Related

• Unsupervised Time-Series Signal Analysis with Autoencoders and Vision Transformers: A Review of Architectures and Applications
• Recurrent Auto-Encoder Model for Large-Scale Industrial Sensor Signal Analysis
• Multivariate Time Series Classification: A Deep Learning Approach
• Continual Learning for Multivariate Time Series Tasks with Variable Input Dimensions
• Handling Variable-Dimensional Time Series with Graph Neural Networks