Multi-Scale Remaining Useful Life Prediction Using Long Short-Term Memory

Round 1

Reviewer 1 Report

The manuscript entitled “Multi-resolution remaining useful life prediction using Long Short-Term Memory” has been investigated in detail. The topic addressed in the manuscript is potentially interesting and the manuscript contains some practical meanings, however, there are some issues which should be addressed by the authors:

1)      In the first place, I would encourage the authors to extend the abstract more with the key results. As it is, the abstract is a little thin and does not quite convey the interesting results that follow in the main paper. The "Abstract" section can be made much more impressive by highlighting your contributions. The contribution of the study should be explained simply and clearly.

2)      The readability and presentation of the study should be further improved. The paper suffers from language problems.

3)      The importance of the design carried out in this manuscript can be explained better than other important studies published in this field. I recommend the authors to review other recently developed works.

4)      What makes the proposed method suitable for this unique task? What new development to the proposed method have the authors added (compared to the existing approaches)? These points should be clarified.

5)      “Discussion” section should be added in a more highlighting, argumentative way. The author should analysis the reason why the tested results is achieved.

6)      The authors should clearly emphasize the contribution of the study. Please note that the up-to-date of references will contribute to the up-to-date of your manuscript. The study named "Detection of solder paste defects with an optimization‐based deep learning model using image processing techniques; Artificial intelligence-based robust hybrid algorithm design and implementation for real-time detection of plant diseases in agricultural environments; Optimization of deep learning model parameters in classification of solder paste defects" - can be used to explain the method in the study or to indicate the contribution in the “Introduction” section.

7)      The complexity of the proposed model and the model parameter uncertainty are not enough mentioned.

8)      The effect of the parametric uncertainty is not discussed in detail. How did the comparison methods perform with or without the uncertainty?

9)      It will be helpful to the readers if some discussions about insight of the main results are added as Remarks.

This study may be proposed for publication if it is addressed in the specified problems.

Author Response

As attached.

Author Response File: Author Response.pdf

Reviewer 2 Report

Broad comments. Using LSTM or other Deep learning methodologies for fault diagnosis is an idea that has been under discussion recently, mainly due to limitations in modeling and theoretical approaches of such complicated dynamical systems.

The authors have made a concise overview of the topic and a brief reference to existing literature. They have indicated the main task of the paper among its motivation. Finally, they have pointed out the key message and the potential benefits of their work. As a general drawback, I could say that there is no reference to similar approaches (e.g. [1]) where the accuracy of machine learning methodologies has been performed on real datasets in vessels.

Theodoropoulos, P.; Spandonidis, C.C.; Themelis, N.; Giordamlis, C.; Fassois, S. Evaluation of Different Deep-Learning Models for the Prediction of a Ship’s Propulsion Power. J. Mar. Sci. Eng. 2021, 9, 116. https://doi.org/10.3390/jmse9020116

Specific comments. In general, the text is very well structured and has clearly defined topics. Some comments for improvement:

1.       While the abstract is well written, the last two sentences should be refined so that the C-Maps dataset is better defined and the outcome of the work is described. For the later, the word flexibility is used which is a bit confusing.

2.       The same as above applies to the introductory section. The authors should better emphasize the impact and novelty of their research, which undoubtedly s high. In addition, the added value of their approach should become more clear early in the paper.

3.       In Table 1 it is not very clear what the numbers in the cells correspond to.

4.       It is not straight why the initial 7 features are discarded in paragraph 3.1.

5.       The authors should justify the selection of operation-based data scaling in paragraph 3.2.

6.       Is Figure 9 needed? Could it be refined to better illustrate the info?

7.       Table 7 includes the outcome of the work. A more detailed discussion on it would be fruitful. Authors should comment on the reduced performance in FD004 for the proposed method

 

Author Response

As attached.

Author Response File: Author Response.pdf

Reviewer 3 Report

The author developed  Multi-resolution remaining useful life prediction using Long Short-Term Memory.

The following queries must be addressed to improve the quality of paper.

1.       The novelty of the work is not properly projected.

2.       Lack of literature survey with recent papers.

3.       Mathematical modelling is not formulated properly. It looks like simple equations.

4.       The strong validation of results is missing

5.       The research objectives stated is not seen in the work and conclusion.

6.       Lack of graphical representation of results.

Author Response

As attached.

Author Response File: Author Response.pdf

Reviewer 4 Report

 

1.     The abstract needs to be restructured, paying attention to logic, and for the validation of prediction results please give information on quantitative indicators.

2.     As for the series of citations in the fifth paragraph of the introduction, I think they lack relevance. What do they want to reveal?

3.     “Multiple resolutions” appear several times in the paper, please explain in detail how multiple resolutions are represented in the engine?

4.     In Figure 3, note the clarity of the image, which will affect the reading, avoid low-level errors, please sift through the full text.

5.     In "RUL target function", " Also, it is more likely to prevent the model from overestimating RUL", please explain the reason for this situation.

6.     I think it is necessary to add references to the analysis and assumptions in section 3.3.

7.     What is the relevance threshold set in "Feature selection"?

8.     In Section 4.3, why was there a relatively significant deviation in the initial phase of the forecast?

9.     I propose to add other evaluation indicators in addition to "RMSE" in the evaluation section.

10.   In Figure 9, please indicate the legend.

11.   I think the authors did not write the advantages of their proposed model in the abstract section. In addition, the presentation of this section is too macroscopic and uncharacteristic, so I suggest embellishing it again.

12.   I think the introduction section lacks a description of the LSTM model, please add this section and suggest citing this paper (Chen et al., 2021).

Chen, Y., He, Y., Zhang, L., Chen, Y., Pu, H., Chen, B., & Gao, L. (2021). Prediction of InSAR deformation time-series using a long short-term memory neural network. International Journal of Remote Sensing, 42(18), 6921–6944. https://doi.org/10.1080/01431161.2021.1947540

13.   can you explain the impact of different sensors from a principle perspective in section 4.1, not just from the result optimization? In section 4.1, when performing feature selection, the correlation method suggests citing relevant literature descriptions. For example, this article can be used as a reference (Fang, S., 2022).

Fang, S., Zhao, Y., Chao, Z. et al. Spatial Distribution Characteristics and Influencing Factors of Tibetan Buddhist Monasteries in Amdo Tibetan Inhabited Regions, China. Journal of Geovisualization & Spatial Analysis, 2022,6(2):29. https://doi.org/10.1007/s41651-022-00124-y

14.   In the result section, I think you can introduce some reference papers to demonstrate the reliability of the results. For example, in sections 4.2, and 4.3, relevant methodological papers should be introduced to support the analysis.

15.   As for the stage separation proposed in this paper, is the more detailed the separation, the better? What costs will the stage separation bring?

Author Response

As attached.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

My comments have been thoroughly addressed. It is acceptable in the present form.

Author Response

Thanks for your acceptance. 

Reviewer 4 Report

The authors introduced machine learning (such as SVM) in the third sentence of the fourth paragraph. However, the authors did not explain its shortcomings in data processing. The following literatures may be helpful for you, it is suggested to add the following references in introduction:

1. Comparing Pan-sharpened Landsat-9 and Sentinel-2 for Land-Use Classification Using Machine Learning Classifiers.

2. Advances of Four Machine Learning Methods for Spatial Data Handling: a Review.

Author Response

Please find the attachment.

Author Response File: Author Response.pdf