An Applicable Predictive Maintenance Framework for the Absence of Run-to-Failure Data
Abstract
:1. Introduction
- In order to be applicable in the field, we propose an autoencoder (AE) based methodology. This is an algorithm that can be modeled directly from the normal data without the failure data.
- In order to apply various models in the future, we divided the steps of making HI and of predicting RUL in the proposed framework. This is the standard applied in recent PdM studies [20].
- The proposed framework is applied to real data cases, not simulation data, to prove the practicality and feasibility of the proposed methodology.
2. Background
2.1. Autoencoder (AE)
2.2. Regression
3. Proposed Framework
3.1. Acquisition and Preprocessing of Data
3.2. Building a Model for HI
- As mentioned in Section 2, it can be calculated by defining the HI with the MAE of the input and output data. Observe the calculated HI which is based on the result of the trained autoencoder. This is to guarantee how the HI is normal and to generate the initial threshold. In general, the HI for the train data will usually be small and uniform if the training is done well.
- The determination of an initial threshold is very challenging. If there is historical run-to-failure data, the initial threshold can be set using the failure data [13]. However, if there is no fault history, the exact threshold is unknown. So, in general, it mostly assumes an arbitrary threshold [6]. The initial threshold can be provided by an expert or determined based on the HI calculated from the training data. In this paper, it is proposed to use a gaussian distribution-based value because z-normalization was applied to the preprocessing method. The general manufacturing process manages each variable with 3-sigma based on the process control method [27]. That is, it is heuristically proposed that the threshold is 3 based on the data normalization and the manufacturing process control method. For example, if the 2-sigma method is utilized, we can use 2 as the threshold. Therefore, it is a method that can be used even if run-to-failure data do not exist, and it is expected that the threshold can be updated when run-to-failure data accumulates. In this case, we use the most common 3-sigma method in process control; a value of 3 can be used as a threshold.
- Whenever new data are collected from the equipment, it can be preprocessed based on the training data, and HI can be calculated using the trained autoencoder. In other words, preprocessing is based on the mean and standard deviation of the training data. If there are some outliers or noise in the new data, a high value of HI can be made despite the normal condition of the equipment. Therefore, it is necessary to focus on the overall trend rather than focusing on each HI value.
3.3. Predict the Remaining Useful Life (RUL)
4. Case Study 1
4.1. Data Description
4.2. Experimental Design
4.3. Experimental Result
5. Case Study 2
5.1. Data Description
5.2. Experimental Design
5.3. Experimental Result
6. Additional Experiment
7. Conclusions
Author Contributions
Funding
Conflicts of Interest
References
- Prieto, L.P.; Rodríguez-Triana, M.J.; Kusmin, M.; Laanpere, M. Smart school multimodal dataset and challenges. CEUR Workshop Proc. 2017, 1828, 53–59. [Google Scholar] [CrossRef]
- Liu, Y.C.; Chang, Y.J.; Liu, S.L.; Chen, S.P. Data-driven prognostics of remaining useful life for milling machine cutting tools. In Proceedings of the 2019 IEEE International Conference on Prognostics and Health Management (ICPHM), San Francisco, CA, USA, 17–20 June 2019; pp. 1–5. [Google Scholar] [CrossRef]
- Heng, A.; Zhang, S.; Tan, A.C.C.; Mathew, J. Rotating machinery prognostics: State of the art, challenges and opportunities. Mech. Syst. Signal Process. 2009, 23, 724–739. [Google Scholar] [CrossRef]
- Aggarwal, K.; Atan, O.; Farahat, A.K.; Zhang, C.; Ristovski, K.; Gupta, C. Two Birds with One Network: Unifying Failure Event Prediction and Time-to-failure Modeling. In Proceedings of the 2018 IEEE International Conference on Big Data (Big Data), Los Angeles, CA, USA, 9–12 December 2019; pp. 1308–1317. [Google Scholar] [CrossRef] [Green Version]
- Zheng, S.; Ristovski, K.; Farahat, A.; Gupta, C. Long Short-Term Memory Network for Remaining Useful Life estimation. In Proceedings of the 2017 IEEE International Conference on Prognostics and Health Management (ICPHM), Dallas, TX, USA, 19–21 June 2017; pp. 88–95. [Google Scholar] [CrossRef]
- Yoo, Y.; Baek, J. A novel image feature for the remaining useful lifetime prediction of bearings based on continuous wavelet transform and convolutional neural network. Appl. Sci. 2018, 8, 1102. [Google Scholar] [CrossRef] [Green Version]
- Zhong, S.; Fu, S.; Lin, L.; Fu, X.; Cui, Z.; Wang, R. A novel unsupervised anomaly detection for gas turbine using Isolation Forest. In Proceedings of the 2019 IEEE International Conference on Prognostics and Health Management (ICPHM), San Francisco, CA, USA, 17–20 June 2019; pp. 1–6. [Google Scholar] [CrossRef]
- Jin, B.; Chen, Y.; Li, D.; Poolla, K.; Sangiovanni-Vincentelli, A. A one-class support vector machine calibration method for time series change point detection. In Proceedings of the 2019 IEEE International Conference on Prognostics and Health Management (ICPHM), San Francisco, CA, USA, 17–20 June 2019; pp. 1–5. [Google Scholar] [CrossRef] [Green Version]
- Zhao, G.; Zhang, G.; Liu, Y.; Zhang, B.; Hu, C. Lithium-ion battery remaining useful life prediction with Deep Belief Network and Relevance Vector Machine. In Proceedings of the 2017 IEEE International Conference on Prognostics and Health Management (ICPHM), Dallas, TX, USA, 19–21 June 2017; pp. 7–13. [Google Scholar] [CrossRef]
- Yao, Q.; Yang, T.; Liu, Z.; Zheng, Z. Remaining useful life estimation by empirical mode decomposition and ensemble deep convolution neural networks. In Proceedings of the 2019 IEEE International Conference on Prognostics and Health Management (ICPHM), San Francisco, CA, USA, 17–20 June 2019; pp. 1–6. [Google Scholar] [CrossRef]
- Si, X.S.; Wang, W.; Hu, C.H.; Zhou, D.H. Remaining useful life estimation—A review on the statistical data driven approaches. Eur. J. Oper. Res. 2011, 213, 1–14. [Google Scholar] [CrossRef]
- Ma, M.; Mao, Z. Deep recurrent convolutional neural network for remaining useful life prediction. In Proceedings of the 2019 IEEE International Conference on Prognostics and Health Management (ICPHM), San Francisco, CA, USA, 17–20 June 2019; pp. 1–4. [Google Scholar] [CrossRef]
- Lin, P.; Tao, J. A novel bearing health indicator construction method based on ensemble stacked autoencoder. In Proceedings of the 2019 IEEE International Conference on Prognostics and Health Management, San Francisco, CA, USA, 17–20 June 2019; pp. 1–9. [Google Scholar] [CrossRef]
- Malhotra, P.; TV, V.; Ramakrishnan, A.; Anand, G.; Vig, L.; Agarwal, P.; Shroff, G. Multi-Sensor Prognostics using an Unsupervised Health Index based on LSTM Encoder-Decoder. arXiv 2016, arXiv:1608.06154. [Google Scholar] [CrossRef]
- Hu, C.; Youn, B.D.; Wang, P.; Taek Yoon, J. Ensemble of data-driven prognostic algorithms for robust prediction of remaining useful life. Reliab. Eng. Syst. Saf. 2012, 103, 120–135. [Google Scholar] [CrossRef] [Green Version]
- Shakya, P.; Kulkarni, M.S.; Darpe, A.K. A novel methodology for online detection of bearing health status for naturally progressing defect. J. Sound Vib. 2014, 333, 5614–5629. [Google Scholar] [CrossRef]
- Yen, F.; Katrib, Z. WaveNet based Autoencoder Model: Vibration Analysis on Centrifugal Pump for Degradation Estimation. In Proceedings of the Annual Conference of the PHM Society, Online, 9–13 November 2020; pp. 1–6. [Google Scholar]
- Li, J.; Wang, L.; Li, Y. Diagnosis method for hydro-generator rotor fault based on stochastic resonance. In Proceedings of the 2019 IEEE International Conference on Prognostics and Health Management (ICPHM), San Francisco, CA, USA, 17–20 June 2019; pp. 1–5. [Google Scholar] [CrossRef]
- Liang, P.; Deng, C.; Wu, J.; Yang, Z.; Zhu, J. Intelligent fault diagnosis of rolling element bearing based on convolutional neural network and frequency spectrograms. In Proceedings of the 2019 IEEE International Conference on Prognostics and Health Management (ICPHM), San Francisco, CA, USA, 17–20 June 2019; pp. 1–5. [Google Scholar] [CrossRef]
- Lei, Y.; Li, N.; Guo, L.; Li, N.; Yan, T.; Lin, J. Machinery health prognostics: A systematic review from data acquisition to RUL prediction. Mech. Syst. Signal Process. 2018, 104, 799–834. [Google Scholar] [CrossRef]
- An, J.; Cho, S. Variational Autoencoder based Anomaly Detection using Reconstruction Probability. Tech. Rep. 2015, 21, 1–18. [Google Scholar]
- Sato, S.; Sanda, K. Degradation estimation of turbines in wind farm using denoising autoencoder model. In Proceedings of the 2019 IEEE International Conference on Prognostics and Health Management (ICPHM), San Francisco, CA, USA, 17–20 June 2019; pp. 1–6. [Google Scholar] [CrossRef]
- Khan, S.; Yairi, T. A review on the application of deep learning in system health management. Mech. Syst. Signal Process. 2018, 107, 241–265. [Google Scholar] [CrossRef]
- Kan, M.S.; Tan, A.C.C.; Mathew, J. A review on prognostic techniques for non-stationary and non-linear rotating systems. Mech. Syst. Signal Process. 2015, 62, 1–20. [Google Scholar] [CrossRef]
- Zhao, R.; Yan, R.; Chen, Z.; Mao, K.; Wang, P.; Gao, R.X. Deep Learning and Its Applications to Machine Health Monitoring: A Survey. Mech. Syst. Signal Process. 2016, 14, 1–14. [Google Scholar] [CrossRef]
- Reddy, K.K.; Sarkar, S.; Venugopalan, V.; Giering, M. Anomaly Detection and Fault Disambiguation in Large Flight Data: A Multi-modal Deep Auto-Encoder Approach; Annual Conference of the Prognostics and Health Management Society: Denver, CO, USA, 2016; pp. 192–199. [Google Scholar]
- Montgomery, D.C. Introduction to Statistical Quality Control, 6th ed.; Wiley: Manitoba, CA, USA, 2008; ISBN 978-0470169926. [Google Scholar]
- Yoon, H.; Park, C.S.; Kim, J.S.; Baek, J.G. Algorithm learning based neural network integrating feature selection and classification. Expert Syst. Appl. 2013, 40, 231–241. [Google Scholar] [CrossRef]
- Kim, D.; Park, S.H.; Baek, J. A Kernel Fisher Discriminant Analysis-based Tree Ensemble Classifier: KFDA Forest. Int. J. Ind. Eng. Theory Appl. Pract. 2018, 25, 569–579. [Google Scholar]
- Lee, J.D.; Li, W.C.; Shen, J.H.; Chuang, C.W. Multi-robotic arms automated production line. In Proceedings of the 2018 4th International Conference on Control, Automation and Robotics, Auckland, New Zealand, 20–23 April 2018; pp. 26–30. [Google Scholar] [CrossRef]
- Wu, T.; Zhang, Y.J.A.; Tang, X. Isolation Forest Based Method for Low-Quality Synchrophasor Measurements and Early Events Detection. In Proceedings of the 2018 IEEE International Conference on Communications, Control, and Computing Technologies for Smart Grids (Smart-GridComm), Aalborg, Denmark, 29–31 October 2018; pp. 1–7. [Google Scholar] [CrossRef]
- Staerman, G.; Mozharovskyi, P.; Clémençon, S.; d’Alché-Buc, F. Functional Isolation Forest. In Proceedings of the Asian Conference on Machine Learning, Nagoya, Japan, 17–19 November 2019; pp. 1–33. [Google Scholar]
- Aremu, O.O.; Cody, R.A.; Hyland-Wood, D.; McAree, P.R. A relative entropy based feature selection framework for asset data in Predictive Maintenance. Comput. Ind. Eng. 2020, 145, 106536. [Google Scholar] [CrossRef]
- Song, Y.; Shi, G.; Chen, L.; Huang, X.; Xia, T. Remaining Useful Life Prediction of Turbofan Engine Using Hybrid Model Based on Autoencoder and Bidirectional Long Short-Term Memory. J. Shanghai Jiaotong Univ. 2018, 23, 85–94. [Google Scholar] [CrossRef]
- Li, X.; Ding, Q.; Sun, J.Q. Remaining useful life estimation in prognostics using deep convolution neural networks. Reliab. Eng. Syst. Saf. 2018, 172, 1–11. [Google Scholar] [CrossRef] [Green Version]
No. | Size of Data (Time Span) | Number of Variables | Fault Class | Name of Equipment |
---|---|---|---|---|
1 | 113 points (about 4 months) | 20 | N1 | K1 |
2 | 258 points (about 8 months) | 20 | N3 | K1 |
3 | 289 points (about 9 months) | 20 | N2 | K2 |
4 | 134 points (about 4 months) | 20 | N3 | K2 |
5 | 84 points (about 3 months) | 20 | N1 | K2 |
Pump Case 1 | Pump Case 2 | Robot Arm Case 1 | Robot Arm Case 2 | Robot Arm Case 3 | Robot Arm Case 4 | Robot Arm Case 5 | ||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Method | RMSE | MAE | RMSE | MAE | RMSE | MAE | RMSE | MAE | RMSE | MAE | RMSE | MAE | RMSE | MAE |
Proposed method | 378.67 | 158.07 | 47.64 | 15.72 | 92.08 | 68.82 | 78.32 | 57.75 | 116.89 | 71.59 | 336.05 | 90.90 | 20.65 | 5.01 |
Isolation forest | 888.69 | 509.43 | 42.38 | 37.54 | 231.30 | 221.70 | 84.23 | 67.35 | 120.89 | 111.29 | 456.76 | 206.77 | 278.68 | 125 |
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. |
© 2021 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/).
Share and Cite
Kim, D.; Lee, S.; Kim, D. An Applicable Predictive Maintenance Framework for the Absence of Run-to-Failure Data. Appl. Sci. 2021, 11, 5180. https://doi.org/10.3390/app11115180
Kim D, Lee S, Kim D. An Applicable Predictive Maintenance Framework for the Absence of Run-to-Failure Data. Applied Sciences. 2021; 11(11):5180. https://doi.org/10.3390/app11115180
Chicago/Turabian StyleKim, Donghwan, Seungchul Lee, and Daeyoung Kim. 2021. "An Applicable Predictive Maintenance Framework for the Absence of Run-to-Failure Data" Applied Sciences 11, no. 11: 5180. https://doi.org/10.3390/app11115180