Transfer Learning for Machine Diagnosis and Prognosis - Revision history

YuantaoFan at 18:32, 6 October 2020

2020-10-06T18:32:30Z

YuantaoFan at 17:11, 4 October 2020

2020-10-04T17:11:32Z

YuantaoFan at 17:00, 4 October 2020

2020-10-04T17:00:11Z

YuantaoFan at 16:25, 4 October 2020

2020-10-04T16:25:43Z

YuantaoFan at 16:24, 4 October 2020

2020-10-04T16:24:31Z

YuantaoFan at 16:24, 4 October 2020

2020-10-04T16:24:03Z

YuantaoFan: Created page with "{{StudentProjectTemplate |Summary=Study and develop deep adversarial neural networks (DANN) based methods to detect faults and predict failures and estimate the health status ..."

2020-10-04T16:23:21Z

Created page with "{{StudentProjectTemplate |Summary=Study and develop deep adversarial neural networks (DANN) based methods to detect faults and predict failures and estimate the health status ..."

New page

{{StudentProjectTemplate
|Summary=Study and develop deep adversarial neural networks (DANN) based methods to detect faults and predict failures and estimate the health status of industrial equipment, under transfer learning settings.
|Keywords=Transfer Learning, Domain adaptation, Domain Adversarial Neural Networks, Fault Diagnosis, Prognosis
|TimeFrame=Fall 2020
|References=[1] Pan, Sinno Jialin, and Qiang Yang. "A survey on transfer learning." IEEE Transactions on knowledge and data engineering 22.10 (2009): 1345-1359.
[2] Zhuang, Fuzhen, et al. "A comprehensive survey on transfer learning." Proceedings of the IEEE (2020).
[3] Ganin, Yaroslav, and Victor Lempitsky. "Unsupervised domain adaptation by backpropagation." arXiv preprint arXiv:1409.7495 (2014).
[4] W. Lu, B. Liang, Y. Cheng, D. Meng, J. Yang, T. Zhang, Deep model-based domain adaptation for fault diagnosis, IEEE Transactions on Industrial Electronics 64 (2017) 2296–2305.
[5] Guo, Liang, et al. "Deep convolutional transfer learning network: A new method for intelligent fault diagnosis of machines with unlabeled data." IEEE Transactions on Industrial Electronics 66.9 (2018): 7316-7325.
[6] Wang, Qin, Gabriel Michau, and Olga Fink. "Domain adaptive transfer learning for fault diagnosis." 2019 Prognostics and System Health Management Conference (PHM-Paris). IEEE, 2019.
[7] Da Costa, Paulo R. de O., et al. "Remaining useful lifetime prediction via deep domain adaptation." arXiv preprint arXiv:1907.07480 (2019).
[8] Akuzawa, Kei, Yusuke Iwasawa, and Yutaka Matsuo. "Adversarial Invariant Feature Learning with Accuracy Constraint for Domain Generalization." arXiv preprint arXiv:1904.12543 (2019).
[9] Long, Mingsheng, et al. "Unsupervised domain adaptation with residual transfer networks." Advances in Neural Information Processing Systems. 2016.
[10] Fan, Yuantao, Sławomir Nowaczyk, and Thorsteinn Rögnvaldsson. "Transfer learning for remaining useful life prediction based on consensus self-organizing models." Reliability Engineering & System Safety 203 (2020): 107098.
[11] Zheng, Huailiang, et al. "Cross-domain fault diagnosis using knowledge transfer strategy: A review." IEEE Access 7 (2019): 129260-129290.
|Prerequisites=Artificial Intelligence, Data Mining, and Learning Systems courses; good knowledge of machine learning and neural networks; programming skills for implementing machine learning algorithms
|Supervisor=Peyman Mashhadi, Yuantao Fan
|Level=Master
|Status=Open
}}
Transfer Learning (TL) [1, 2, 11] refers to methods for transferring knowledge learned in one setting (the source domain) to another setting (the target domain) and it is needed in many fields, especially in the application area of machine diagnostics and prognostics. The current industrial approach for developing machine diagnostic and prognostic methods usually relies on data acquired in experiments under controlled conditions prior to deployment of the equipment. Detecting faults, predicting failures, and estimating machine health in this paradigm assumes that future field data will have a very similar distribution to the experiment data. However, many machines were operated under changing/evolving environmental conditions and were operated in a variety of ways. This reply on the assumption that pre-deployment data and post-deployment data follow a very similar distribution is unlikely to hold. Moreover, since industrial systems are not allowed to run until failure (for safety reasons), collecting data that has comprehensive coverage on various usages patterns, fault types and deterioration patterns are very challenging. Utilizing available data (including fault and failure cases) from other equipment that shares a similar mechanical structure and/or being operated in a similar way with adaptation will be helpful if done properly.

Therefore, machinery diagnosis and prognosis need to: (i) adapt to more complex scenarios where unseen degradation patterns and new operating conditions are present (in the testing data); (ii) learn to utilize and transfer knowledge gained from operations of similar equipment/assets. A suitable solution is to perform feature-representation-based TL (or domain adaptation) methods, for transferring knowledge between tasks, e.g. [6], and/or dealing with new conditions/faults, e.g. [10]. Feature-representation-based TL and domain adaptation methods aim at discovering meaningful common structures between the source and the target domain, finding transformations that project the source data and the target data into a common latent feature space, which has predictive qualities for solving the target task. The discrepancy of marginal distributions between the source and the target data in the latent feature space is expected to be reduced at the same time. Domain Adversarial Neural Networks (DANN) [3, 9] has been applied for fault diagnosis and machine prognosis [4, 5, 6, 7, 8]. The idea is to train a deep neural network for extracting domain-invariant features that has predictive power for the classification/regression tasks.

The main objective of this work is to develop a DANN based method, e.g. designing network structure and cost functions, etc., to perform machine diagnosis and prognosis, under transfer learning scenarios. The proposed method will be evaluated using both simulated data and real data from industrial systems.

@@ Line 30: / Line 30: @@
 |Status=Open
 }}
-Transfer Learning (TL) [1, 2, 11] refers to methods for transferring knowledge learned in one setting (the source domain) to another setting (the target domain) and it is needed in many fields, especially in the application area of machine diagnosis and prognosis. The current industrial approach for developing machine diagnosis and prognosis methods usually relies on data acquired in experiments under controlled conditions prior to deployment of the equipment. Detecting faults, predicting failures, and estimating machine health in this paradigm assumes that future field data will have a very similar distribution to the experiment data. However, many machines were operated under changing/evolving environmental conditions and were operated in a variety of ways. This reply on the assumption that pre-deployment data and post-deployment data follow a very similar distribution is unlikely to hold. Moreover, since industrial systems are not allowed to run until failure (for safety reasons), collecting data that has comprehensive coverage on various usages patterns, fault types and deterioration patterns are very challenging. Utilizing available data (including fault and failure cases) from other equipment that shares a similar mechanical structure and/or being operated in a similar way with adaptation will be helpful if done properly.
+Transfer Learning (TL) [1, 2, 11] refers to methods for transferring knowledge learned in one setting (the source domain) to another setting (the target domain) and it is needed in many fields, especially in the application area of machine diagnosis and prognosis. The current industrial approach for machine diagnosis and prognosis usually relies on data acquired in experiments under controlled conditions prior to deployment of the equipment, which might not represent the (operating/environment) conditions after being deployed to the real-world application. Moreover, since industrial systems are not allowed to run until failure (for safety reasons), collecting data that has comprehensive coverage on usages patterns, faults, and equipment deterioration patterns are very challenging. Utilizing available data (including fault and failure cases) from other equipment that shares a similar mechanical structure and/or being operated in a similar way with adaptation will be helpful if done properly.
-Therefore, machinery diagnosis and prognosis need to: (i) adapt to more complex scenarios where unseen degradation patterns and new operating conditions are present (in the testing data); (ii) learn to utilize and transfer knowledge gained from operations of similar equipment/assets. A suitable solution is to perform feature-representation-based TL (or domain adaptation) methods, for transferring knowledge between tasks, e.g. [6], and/or dealing with new conditions/faults, e.g. [10]. Feature-representation-based TL and domain adaptation methods aim at discovering meaningful common structures between the source and the target domain, finding transformations that project the source data and the target data into a common latent feature space, which has predictive qualities for solving the target task. The discrepancy of marginal distributions between the source and the target data in the latent feature space is expected to be reduced at the same time. Domain Adversarial Neural Networks (DANN) [3, 9] has been applied for fault diagnosis and machine prognosis [4, 5, 6, 7, 8]. The idea is to train a deep neural network for extracting domain-invariant features that has predictive power for the classification/regression tasks.
+Therefore, machine diagnosis and prognosis need to: (i) adapt to more complex scenarios where unseen degradation patterns and new operating conditions are present (in the testing data); (ii) learn to utilize and transfer knowledge gained from operations of similar equipment/assets. A suitable solution is to apply transfer learning (or domain adaptation) methods, utilize and transfer knowledge between different tasks, e.g. [6], and/or dealing with new conditions/faults, e.g. [10]. As a popular feature-representation-based TL method, Domain Adversarial Neural Networks (DANN) [3, 9] has been applied for fault diagnosis and machine prognosis [4, 5, 6, 7, 8]. The idea is to train a deep neural network for extracting domain-invariant features that has predictive power for the classification/regression tasks.
-The main objective of this work is to develop a DANN based method, e.g. designing network structure and cost functions, etc., to perform machine diagnosis and prognosis, under transfer learning scenarios. The thesis are expected to address the generality of DANN based approaches in dealing with different types of transfer learning scenarios. The proposed method will be evaluated using simulated data and/or real data from industrial systems.
+The main objective of this work is to develop a DANN based method, e.g. designing network structure and cost functions, etc., to perform machine diagnosis and prognosis, under transfer learning scenarios. The thesis is expected to address the generality of DANN based approaches in dealing with different types of transfer learning scenarios. The proposed method will be evaluated using simulated data and/or real data from industrial systems.

← Older revision		Revision as of 17:00, 4 October 2020
Line 34:		Line 34:
	Therefore, machinery diagnosis and prognosis need to: (i) adapt to more complex scenarios where unseen degradation patterns and new operating conditions are present (in the testing data); (ii) learn to utilize and transfer knowledge gained from operations of similar equipment/assets. A suitable solution is to perform feature-representation-based TL (or domain adaptation) methods, for transferring knowledge between tasks, e.g. [6], and/or dealing with new conditions/faults, e.g. [10]. Feature-representation-based TL and domain adaptation methods aim at discovering meaningful common structures between the source and the target domain, finding transformations that project the source data and the target data into a common latent feature space, which has predictive qualities for solving the target task. The discrepancy of marginal distributions between the source and the target data in the latent feature space is expected to be reduced at the same time. Domain Adversarial Neural Networks (DANN) [3, 9] has been applied for fault diagnosis and machine prognosis [4, 5, 6, 7, 8]. The idea is to train a deep neural network for extracting domain-invariant features that has predictive power for the classification/regression tasks.		Therefore, machinery diagnosis and prognosis need to: (i) adapt to more complex scenarios where unseen degradation patterns and new operating conditions are present (in the testing data); (ii) learn to utilize and transfer knowledge gained from operations of similar equipment/assets. A suitable solution is to perform feature-representation-based TL (or domain adaptation) methods, for transferring knowledge between tasks, e.g. [6], and/or dealing with new conditions/faults, e.g. [10]. Feature-representation-based TL and domain adaptation methods aim at discovering meaningful common structures between the source and the target domain, finding transformations that project the source data and the target data into a common latent feature space, which has predictive qualities for solving the target task. The discrepancy of marginal distributions between the source and the target data in the latent feature space is expected to be reduced at the same time. Domain Adversarial Neural Networks (DANN) [3, 9] has been applied for fault diagnosis and machine prognosis [4, 5, 6, 7, 8]. The idea is to train a deep neural network for extracting domain-invariant features that has predictive power for the classification/regression tasks.

−	The main objective of this work is to develop a DANN based method, e.g. designing network structure and cost functions, etc., to perform machine diagnosis and prognosis, under transfer learning scenarios. The proposed method will be evaluated using ~~both~~ simulated data and real data from industrial systems.	+	The main objective of this work is to develop a DANN based method, e.g. designing network structure and cost functions, etc., to perform machine diagnosis and prognosis, under transfer learning scenarios. The thesis are expected to address the generality of DANN based approaches in dealing with different types of transfer learning scenarios. The proposed method will be evaluated using simulated data and/or real data from industrial systems.

@@ Line 26: / Line 26: @@
 [11] Zheng, Huailiang, et al. "Cross-domain fault diagnosis using knowledge transfer strategy: A review." IEEE Access 7 (2019): 129260-129290.
 |Prerequisites=Artificial Intelligence, Data Mining, and Learning Systems courses; good knowledge of machine learning and neural networks; programming skills for implementing machine learning algorithms
-|Supervisor=Peyman Mashhadi, Yuantao Fan
+|Supervisor=Peyman Mashhadi, Yuantao Fan, Mohammed Ghaith Altarabichi
 |Level=Master
 |Status=Open

@@ Line 4: / Line 4: @@
 |TimeFrame=Fall 2020
 |References=[1] Pan, Sinno Jialin, and Qiang Yang. "A survey on transfer learning." IEEE Transactions on knowledge and data engineering 22.10 (2009): 1345-1359.
 [2] Zhuang, Fuzhen, et al. "A comprehensive survey on transfer learning." Proceedings of the IEEE (2020).
 [3] Ganin, Yaroslav, and Victor Lempitsky. "Unsupervised domain adaptation by backpropagation." arXiv preprint arXiv:1409.7495 (2014).
 [4] W. Lu, B. Liang, Y. Cheng, D. Meng, J. Yang, T. Zhang, Deep model-based domain adaptation for fault diagnosis, IEEE Transactions on Industrial Electronics 64 (2017) 2296–2305.
 [5] Guo, Liang, et al. "Deep convolutional transfer learning network: A new method for intelligent fault diagnosis of machines with unlabeled data." IEEE Transactions on Industrial Electronics 66.9 (2018): 7316-7325.
 [6] Wang, Qin, Gabriel Michau, and Olga Fink. "Domain adaptive transfer learning for fault diagnosis." 2019 Prognostics and System Health Management Conference (PHM-Paris). IEEE, 2019.
 [7] Da Costa, Paulo R. de O., et al. "Remaining useful lifetime prediction via deep domain adaptation." arXiv preprint arXiv:1907.07480 (2019).
 [8] Akuzawa, Kei, Yusuke Iwasawa, and Yutaka Matsuo. "Adversarial Invariant Feature Learning with Accuracy Constraint for Domain Generalization." arXiv preprint arXiv:1904.12543 (2019).
 [9] Long, Mingsheng, et al. "Unsupervised domain adaptation with residual transfer networks." Advances in Neural Information Processing Systems. 2016.
 [10] Fan, Yuantao, Sławomir Nowaczyk, and Thorsteinn Rögnvaldsson. "Transfer learning for remaining useful life prediction based on consensus self-organizing models." Reliability Engineering & System Safety 203 (2020): 107098.
 [11] Zheng, Huailiang, et al. "Cross-domain fault diagnosis using knowledge transfer strategy: A review." IEEE Access 7 (2019): 129260-129290.
 |Prerequisites=Artificial Intelligence, Data Mining, and Learning Systems courses; good knowledge of machine learning and neural networks; programming skills for implementing machine learning algorithms

@@ Line 1: / Line 1: @@
 {{StudentProjectTemplate
-|Summary=Study and develop deep adversarial neural networks (DANN) based methods to detect faults and predict failures and estimate the health status of industrial equipment, under transfer learning settings.
+|Summary=Study and develop deep adversarial neural networks (DANN) based methods to detect faults and predict failures in industrial equipment, under transfer learning scenarios.
 |Keywords=Transfer Learning, Domain adaptation, Domain Adversarial Neural Networks, Fault Diagnosis, Prognosis
 |TimeFrame=Fall 2020