A Novel Deep Clustering Method and Indicator for Time Series Soft Partitioning
The aerospace industry develops prognosis and health management algorithms to ensure better safety on board, particularly for in-flight controls where jamming is dreaded. For that, vibration signals are monitored to predict future defect occurrences. However, time series are not labeled according to...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2021-09-01
|
| Series: | Energies |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1996-1073/14/17/5530 |
| _version_ | 1797521482234986496 |
|---|---|
| author | Alexandre Eid Guy Clerc Badr Mansouri Stella Roux |
| author_facet | Alexandre Eid Guy Clerc Badr Mansouri Stella Roux |
| author_sort | Alexandre Eid |
| collection | DOAJ |
| description | The aerospace industry develops prognosis and health management algorithms to ensure better safety on board, particularly for in-flight controls where jamming is dreaded. For that, vibration signals are monitored to predict future defect occurrences. However, time series are not labeled according to severity level, and the user can only assess the system health from the data mining procedure. To that extent, a clustering algorithm using a deep neural network core is developed. Time series are encoded into pictures to be fed into an artificially trained neural network: U-NET. From the segmented output, one-dimensional information on cluster frontiers is extracted and filtered without any parameter selection. Then, a kernel density estimation finally transforms the signal into an empirical density. Ultimately, a Gaussian mixture model extracts the latter independent components. The method empowered us to reveal different degrees of severity faults in the studied data, with their respective likelihoods, without prior knowledge. It was then compared to state-of-the-art machine learning algorithms. However, internal clustering results evaluation for time series is an open question. As the state-of-the-art indexes were not producing relevant results, a new indicator was built to fulfill this task. We applied the whole method to an actuator consisting of an induction machine linked to a ball screw. This study lays the groundwork for future training of diagnosis and prognosis structures in the health management framework. |
| first_indexed | 2024-03-10T08:12:12Z |
| format | Article |
| id | doaj.art-ea284f05f8f54656b4854798c5d04b29 |
| institution | Directory Open Access Journal |
| issn | 1996-1073 |
| language | English |
| last_indexed | 2024-03-10T08:12:12Z |
| publishDate | 2021-09-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Energies |
| spelling | doaj.art-ea284f05f8f54656b4854798c5d04b292023-11-22T10:36:04ZengMDPI AGEnergies1996-10732021-09-011417553010.3390/en14175530A Novel Deep Clustering Method and Indicator for Time Series Soft PartitioningAlexandre Eid0Guy Clerc1Badr Mansouri2Stella Roux3University Lyon, Université Claude Bernard Lyon 1, INSA Lyon, École Centrale de Lyon CNRS, Ampère, UMR5005, 69622 Villeurbanne, FranceUniversity Lyon, Université Claude Bernard Lyon 1, INSA Lyon, École Centrale de Lyon CNRS, Ampère, UMR5005, 69622 Villeurbanne, FranceSafran Electronics & Defense, 91344 Massy, FranceGrenoble INP—Ensimag, UGA, 38400 Saint-Martin-d’Hères, FranceThe aerospace industry develops prognosis and health management algorithms to ensure better safety on board, particularly for in-flight controls where jamming is dreaded. For that, vibration signals are monitored to predict future defect occurrences. However, time series are not labeled according to severity level, and the user can only assess the system health from the data mining procedure. To that extent, a clustering algorithm using a deep neural network core is developed. Time series are encoded into pictures to be fed into an artificially trained neural network: U-NET. From the segmented output, one-dimensional information on cluster frontiers is extracted and filtered without any parameter selection. Then, a kernel density estimation finally transforms the signal into an empirical density. Ultimately, a Gaussian mixture model extracts the latter independent components. The method empowered us to reveal different degrees of severity faults in the studied data, with their respective likelihoods, without prior knowledge. It was then compared to state-of-the-art machine learning algorithms. However, internal clustering results evaluation for time series is an open question. As the state-of-the-art indexes were not producing relevant results, a new indicator was built to fulfill this task. We applied the whole method to an actuator consisting of an induction machine linked to a ball screw. This study lays the groundwork for future training of diagnosis and prognosis structures in the health management framework.https://www.mdpi.com/1996-1073/14/17/5530semantic segmentationtime seriesclusteringdeep learningkernel density estimationelectromechanical actuator |
| spellingShingle | Alexandre Eid Guy Clerc Badr Mansouri Stella Roux A Novel Deep Clustering Method and Indicator for Time Series Soft Partitioning Energies semantic segmentation time series clustering deep learning kernel density estimation electromechanical actuator |
| title | A Novel Deep Clustering Method and Indicator for Time Series Soft Partitioning |
| title_full | A Novel Deep Clustering Method and Indicator for Time Series Soft Partitioning |
| title_fullStr | A Novel Deep Clustering Method and Indicator for Time Series Soft Partitioning |
| title_full_unstemmed | A Novel Deep Clustering Method and Indicator for Time Series Soft Partitioning |
| title_short | A Novel Deep Clustering Method and Indicator for Time Series Soft Partitioning |
| title_sort | novel deep clustering method and indicator for time series soft partitioning |
| topic | semantic segmentation time series clustering deep learning kernel density estimation electromechanical actuator |
| url | https://www.mdpi.com/1996-1073/14/17/5530 |
| work_keys_str_mv | AT alexandreeid anoveldeepclusteringmethodandindicatorfortimeseriessoftpartitioning AT guyclerc anoveldeepclusteringmethodandindicatorfortimeseriessoftpartitioning AT badrmansouri anoveldeepclusteringmethodandindicatorfortimeseriessoftpartitioning AT stellaroux anoveldeepclusteringmethodandindicatorfortimeseriessoftpartitioning AT alexandreeid noveldeepclusteringmethodandindicatorfortimeseriessoftpartitioning AT guyclerc noveldeepclusteringmethodandindicatorfortimeseriessoftpartitioning AT badrmansouri noveldeepclusteringmethodandindicatorfortimeseriessoftpartitioning AT stellaroux noveldeepclusteringmethodandindicatorfortimeseriessoftpartitioning |