2017
DOI: 10.1016/j.microrel.2017.03.021
|View full text |Cite
|
Sign up to set email alerts
|

EMA remaining useful life prediction with weighted bagging GPR algorithm

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1
1
1

Citation Types

0
9
0

Year Published

2017
2017
2021
2021

Publication Types

Select...
9
1

Relationship

1
9

Authors

Journals

citations
Cited by 40 publications
(9 citation statements)
references
References 40 publications
0
9
0
Order By: Relevance
“…Finally, we use the prediction results on these seven bearings to evaluate the performance of the proposed method. Root mean squared error (RMSE) and mean absolute percentage error (MAPE) are introduced to evaluate the prediction performance with the following formula 31 GPR, 32 SVM, 5 and extreme learning machine (ELM). 33 These four methods also use the same deep features extracted in section ''Deep feature extraction'' to train the regression model and then conduct RUL prediction for the bearing data in fast degradation state.…”
Section: Resultsmentioning
confidence: 99%
“…Finally, we use the prediction results on these seven bearings to evaluate the performance of the proposed method. Root mean squared error (RMSE) and mean absolute percentage error (MAPE) are introduced to evaluate the prediction performance with the following formula 31 GPR, 32 SVM, 5 and extreme learning machine (ELM). 33 These four methods also use the same deep features extracted in section ''Deep feature extraction'' to train the regression model and then conduct RUL prediction for the bearing data in fast degradation state.…”
Section: Resultsmentioning
confidence: 99%
“…For building an RUL prediction model, we choose LIB-SVM [41] as the implementation of SVM. We introduce the other three regression algorithms for comparison: linear regression (LR) [42], Gaussian process regression (GPR) [43] and least square SVM (LSSVM) [44]. For these four regression algorithms, model selection is conducted to choose the optimal hyper-parameters.…”
Section: B Experimental Settingmentioning
confidence: 99%
“…Note that the flatten layer is used as the deep feature representation of the vibration signals to be fed into the LSTM network for RUL prediction. The proposed method is compared with other methods, such as the SVM [45], linear regression (LR) [46], GPR [47], and extreme learning machine (ELM) [48], to show its superiority.…”
Section: Related Workmentioning
confidence: 99%