Predicting Failure Probability in Industry 4.0 Production Systems: A Workload-Based Prognostic Model for Maintenance Planning
Round 1
Reviewer 1 Report
The manuscript is well in the scope of the journal and can be considered as the area of interest for readers. There are few issues in the manuscript which needs to be addressed:
1. Abstract of the study is too general and lacks in novelty. The abstract can be improved by adding how this study can be beneficial for the present research community? How this study can be useful for industries to address present issues related to Industry 4.0? These all issues can be addressed in the abstract in order to improve the abstract.
2. Introduction is well written up to some extent but again taxonomy of Industry 4.0 is not presented well. In my suggestions authors must follow few goods studies from the MDPI journals which will help to develop a good introduction section. The studies are as:
(2021) Industry 4.0 technologies for manufacturing sustainability: a systematic review and future research directions
(2020) Industry 4.0 readiness models: a systematic literature review of model dimensions
(2020) Practical application of the Industry 4.0 concept in a steel company
(2019) Tackling faults in the industry 4.0 era—a survey of machine-learning solutions and key aspects
These are few studies from MDPI journals related to this study which can be added in the present study.
3. No research questions or objective have been added in the study which needs to be added.
4. Literature review strategy needs to be discussed.
5. Figure 1 can be improved by adding what are the physical processes?
6. Figure 3 lacks in clarity can be improved by adding high resolution image.
7. Equations needs to be explained properly.
8. . How Table 1 data is collected?
9. Figure 3 is not clear.
10. Figure 5 needs revision. Figure 8 and Figure 9 also lacks in clarity.
11.Implications of work is not stated.
12. Conclusion is weak and needs to be supported by novel results. Future scopes are too general and can be revised.
13. Paper needs to be formatted in the MDPI template. Reference style also needs to be changed.
The manuscript is good and can be accepted after a major revision.
Author Response
First, we would like to thank the Editor who gave us the opportunity to resubmit a revised version of the paper. Then, we would like to acknowledge the Reviewers for their thoughtful review of the manuscript and for their precious comments, which permitted us to really improve the original version of the paper. We have carefully addressed each observation raised by the Reviewers and the manuscript has been adjusted accordingly. Modifications made to the text according to Reviewers’ comments and suggestions have been appropriately reported in the manuscript. Below, we give the response to each specific comment that has been brought to our attention. We hope we were able to satisfy all concerns and that the paper is now adequate for publication on Applied Sciences.
Comments No 1: “[..] 1. Abstract of the study is too general and lacks in novelty. The abstract can be improved by adding how this study can be beneficial for the present research community? How this study can be useful for industries to address present issues related to Industry 4.0? These all issues can be addressed in the abstract in order to improve the abstract.”
Our response: We would like to thank the Reviewer for this comment, and we have modified the abstract according to her/his valuable modification suggestions.
Comments No 2: “[..] 2. Introduction is well written up to some extent but again taxonomy of Industry 4.0 is not presented well. In my suggestions authors must follow few goods studies from the MDPI journals which will help to develop a good introduction section. The studies are as:
(2021) Industry 4.0 technologies for manufacturing sustainability: a systematic review and future research directions
(2020) Industry 4.0 readiness models: a systematic literature review of model dimensions
(2020) Practical application of the Industry 4.0 concept in a steel company
(2019) Tackling faults in the industry 4.0 era—a survey of machine-learning solutions and key aspects
These are few studies from MDPI journals related to this study which can be added in the present study.”
Our response: We appreciate this comment pertaining to the introduction section as it permits us better establish the domain of interest of the paper as well as to define some reference terms. The section has been radically modified also benefiting of the relevant papers suggested by the Reviewer.
Comments No 3: “[..] 3. No research questions or objective have been added in the study which needs to be added. ”
Our response: We have clearly formulated the objective of the paper in the introduction section.
Comments No 4: “[..] 4. Literature review strategy needs to be discussed.”
Our response: We would like to thank the Reviewer for the comment as it permitted us to discuss the rationale and the criteria according to which we carried out the literature review. This discussion can be found at the beginning of the literature review section (subsection 2.1).
Comments No 5: “[..] 5. Figure 1 can be improved by adding what are the physical processes?”
Our response: Figure 1 has been modified according to this valuable comment.
Comments No 6: “[..] 6. Figure 3 lacks in clarity can be improved by adding high resolution image.”
Our response: We apologize for the low quality of Figure 3. According to this comment, we have added a new high-resolution image.
Comments No 7: “[..] 7. Equations needs to be explained properly.”
Our response: We are sorry for any lack of clarity or exhaustiveness in describing the various equations throughout the manuscript. We went through the whole manuscript, and we have carefully justified the various equations, when necessary.
Comments No 8: “[..] 8. . How Table 1 data is collected?”
Our response: We would like to thank the Reviewer for this comment as it permitted us to clarify some aspects of our research. Specifically, it is worth noting that Table 1 data are fictious being only an example of classification of jobs based on the expected mechanical load.
Comments No 9: “[..] 9. Figure 9 is not clear.”
Our response: We apologize for the low quality of Figure 9. According to this comment, we have added a new high-resolution image.
Comments No 10: “[..] 10. Figure 5 needs revision. Figure 8 and Figure 9 also lacks in clarity.”
Our response: Again, we are sorry for Figures 5, 8, 9. According to this comment, we have modified them to appropriately tackle the issue.
Comments No 11: “[..] 11. Implications of work is not stated.”
Our response: We would like to thank the Reviewer for this comment as it highlighted an aspect of this research that was not addressed. According to this comment, we have added a new paragraph to the manuscript in the Case Study section, where we discuss the results of the case study and, as a result, draw insights and implications of the study.
Comments No 12: “[..]12. Conclusion is weak and needs to be supported by novel results. Future scopes are too general and can be revised.”
Our response: According to this comment, we have revised the conclusive section of the paper to include the novel results brought about by this research. Moreover, always in this section, we have properly addressed potential future developments of this research.
Comments No 13: “[..] 13. Paper needs to be formatted in the MDPI template. Reference style also needs to be changed.”
Our response: We have formatted the paper according to the MPDI template and we have adapted consistently also the reference style.
Comments No 14: “[..] The manuscript is good and can be accepted after a major revision. ”
Our response: Thank you for the positive feedback and we hope to have satisfied all the concerns raised by the Reviewer.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
· The introduction is written clearly, and the paper's contribution is well explained. Yet, to validate the results it's crucial to add a comparison with prior work, even if it doesn’t include the workload to show the actual effect of the proposed work. Otherwise, the paper's contribution is not solid.
· In the introduction section, the paper contributions are replicated, (ii and iii) are the same
· In the literature review, a table summarizing the similar research work done in PHM prognostic would be beneficial as the descriptive text doesn’t show the pros and cons of each research methodology
· Reference numbering is incorrect, please revise and modify
· Some grammar and punctuation mistakes need to be corrected
· Replacing Table 3 with a graph would be more beneficial to show the results of the proposed algorithm
Author Response
First, we would like to thank the Editor who gave us the opportunity to resubmit a revised version of the paper. Then, we would like to acknowledge the Reviewers for their thoughtful review of the manuscript and for their precious comments, which permitted us to really improve the original version of the paper. We have carefully addressed each observation raised by the Reviewers and the manuscript has been adjusted accordingly. Modifications made to the text according to Reviewers’ comments and suggestions have been appropriately reported in the manuscript. Below, we give the response to each specific comment that has been brought to our attention. We hope we were able to satisfy all concerns and that the paper is now adequate for publication on Applied Sciences.
Comments No 1: “[..] The introduction is written clearly, and the paper's contribution is well explained. Yet, to validate the results it's crucial to add a comparison with prior work, even if it doesn’t include the workload to show the actual effect of the proposed work. Otherwise, the paper's contribution is not solid.”
Our response: We would like to thank the Reviewer for her/his positive feedback, in addition, we and we have modified the abstract according to her/his valuable modification suggestions.
Comments No 2: “[..] In the introduction section, the paper contributions are replicated, (ii and iii) are the same.”
Our response: We apologize for this oversight, and we confirm that the manuscript has been amended.
Comments No 3: “[..] In the literature review, a table summarizing the similar research work done in PHM prognostic would be beneficial as the descriptive text doesn’t show the pros and cons of each research methodology”
Our response: Thank you for your valuable comment. We understand the importance of summarizing similar research work in the field of PHM prognostic and appreciate your suggestion to include a table for this purpose. However, we have chosen to maintain an informative style in our literature review and have added a new subsection at the beginning of the literature review paragraph to clarify the methodology used.
Comments No 4: “[..] Reference numbering is incorrect, please revise and modify”
Our response: We would like to thank the Reviewer for her/his careful reading, and we confirm that we have double-checked references and their numbering.
Comments No 5: “[..] Some grammar and punctuation mistakes need to be corrected ”
Our response: We apologise for typos, mistakes, and any other error present in the manuscript. We have gone through all of it making the necessary changes.
Comments No 6: “[..] Replacing Table 3 with a graph would be more beneficial to show the results of the proposed algorithm. ”
Our response: Thanks for your suggestion to replace Table 3 with a graph in order to better display the results of the proposed algorithm. We did explore the option of creating a graph using the data provided in the table, but ultimately found that the representation in the table format was more effective in helping the reader to gauge the prediction error advantages of the proposed methodology.
Author Response File: Author Response.pdf
Reviewer 3 Report
After reading manuscript, there are several modifications required in manuscript, which are as follows :
1. Abstract required addition of two/three lines specifically related to outcomes of results.
2. In page 3, line 95-98,authors mentioned that degradation model is developed using logistic regression model, whereas FNN is used for prognostic model. It is unclear to reader what is the use of two ML algorithms. FNN can be used both for degradation and prognostics.
3. There is no need of separate section 2 as literature review. It should be merged with Introduction section itself.
4. Do there is a need of Figure 1.It is a general figure.It would be much better if Fig.1. is replaced by Flowchart of authors methodology.
5. Resolution of all figures need to be improved.At present, figures seems to be blurred.
6. In pg.19,authors mentioned that dataset was divided in three parts. This is an unreliable approach as during random partition of data in training,testing and validation approach. Unbiased and reliable results can be generated by using full dataset as training and apply 10-fold CV on same dataset. Kindly refer following journals and add in revised manuscript:
a. https://www.mdpi.com/2075-1702/10/3/176
b. https://www.mdpi.com/2075-1702/7/4/74
7. It is always a best practice to compare results with at least two algorithms. Kindly refer 6a.
8. Manuscript seems to be very lengthy with unwanted theoretical descriptions. Manuscript need to be shortened and should more focus on results and analysis portion. Even comparative results should be included which will strengthened the manuscript in technical aspects.
Author Response
First, we would like to thank the Editor who gave us the opportunity to resubmit a revised version of the paper. Then, we would like to acknowledge the Reviewers for their thoughtful review of the manuscript and for their precious comments, which permitted us to really improve the original version of the paper. We have carefully addressed each observation raised by the Reviewers and the manuscript has been adjusted accordingly. Modifications made to the text according to Reviewers’ comments and suggestions have been appropriately reported in the manuscript. Below, we give the response to each specific comment that has been brought to our attention. We hope we were able to satisfy all concerns and that the paper is now adequate for publication on Applied Sciences.
Comments No 1: “[..] 1. Abstract required addition of two/three lines specifically related to outcomes of results. ”
Our response: We would like to thank the Reviewer for this comment, and we have modified the abstract to include the key results of the research.
Comments No 2: “[..] 2. In page 3, line 95-98,authors mentioned that degradation model is developed using logistic regression model, whereas FNN is used for prognostic model. It is unclear to reader what is the use of two ML algorithms. FNN can be used both for degradation and prognostics.”
Our response: Thank you for bringing attention to page 3, line 95-98 in our scientific work. We agree with your observation that it may be unclear to the reader why we have chosen to use two different machine learning algorithms, logistic regression and FNN, for the degradation and prognostic models. As discussed in the Proposal section, we have chosen to use logistic regression for the degradation assessment due to its wide utilization in the literature and its stability. While it is true that FNN can be used for both degradation and prognostics, we believed that using logistic regression for degradation assessment would provide a more robust and reliable model. On the other hand, we have chosen to use FNN for prognostics as it has been shown to be more accurate in predicting the remaining useful life of the system. We apologize if this was not made clear in the original manuscript, and we hope that this explanation clarifies any confusion.
Comments No 3: “[..] 3. There is no need of separate section 2 as literature review. It should be merged with Introduction section itself. ”
Our response: We have long pondered this comment and we think that merging these two sections would result in a too hefty introduction which would endanger a smooth reading of the paper.
Comments No 4: “[..] 4. Do there is a need of Figure 1.It is a general figure.It would be much better if Fig.1. is replaced by Flowchart of authors methodology. ”
Our response: Thank you for your suggestion regarding Figure 1 in our scientific work. However, after considering the feedback from other reviewers, we have improved the detail and information provided by the proposed images, and we believe that Figure 1 is still a valuable contribution for the reader. In regards to the flowchart methodology, we would like to inform you that its already graphically described in Figure 2 manuscript.
Comments No 5: “[..] 5. Resolution of all figures need to be improved. At present, figures seems to be blurred.”
Our response: We apologize for the low resolution of Figures which have been replaced with high resolution ones.
Comments No 6: “[..] 6. In pg.19, authors mentioned that dataset was divided in three parts. This is an unreliable approach as during random partition of data in training, testing and validation approach. Unbiased and reliable results can be generated by using full dataset as training and apply 10-fold CV on same dataset. Kindly refer following journals and add in revised manuscript:
- https://www.mdpi.com/2075-1702/10/3/176
- https://www.mdpi.com/2075-1702/7/4/74.”
Our response: We appreciate the reviewer’s recommendation to use a 10-fold cross validation approach on the full dataset instead of dividing it into three parts for training, testing, and validation. However, we have used the random partition approach for the training of the prognostic model as it is widely used by literature for similar models. To be exhaustive, regarding your comment we have tried the proposed 10-fold cross validation approach on the full dataset and we found that it does not result in a significant improvement in the performance estimation of the model. Nevertheless, we consider the Reviewer’s suggestion a potential future development of this research.
Comments No 7: “[..] 7. It is always a best practice to compare results with at least two algorithms. Kindly refer 6a. ”
Our response: We definitely agree with this comment however we would like to make some observations. Basically, the objective of this research was to develop a prognostic model which could factor in machine-workload information. The performance of the proposed model has been compared to that of a model which does not account for machine-workload. Nevertheless, we consider the Reviewer’s suggestion a potential future development of this research. In fact, a future course of action could be that of changing the ML technique included in the framework to assess which configuration works best.
Comments No 8: “[..] 8. Manuscript seems to be very lengthy with unwanted theoretical descriptions. Manuscript need to be shortened and should more focus on results and analysis portion. Even comparative results should be included which will strengthened the manuscript in technical aspects. ”
Our response: We appreciate this comment and we have shortened Section 1 and Section 2.1 (now Section 2.2) which contains most of theoretical descriptions. As far as comparative results are concerned, please refer to the answer to comment No.7.
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
I do respect the authors' point of view in response to my comments, yet all of them were not considered. I even asked for introduction modification and it was reflected in the abstract.
Without any quantative comparison or clear qualittative comparsion the paper contribution is not clear.
Author Response
Comments No 1: I do respect the authors' point of view in response to my comments, yet all of them were not considered. I even asked for introduction modification and it was reflected in the abstract. Without any quantative comparison or clear qualittative comparsion the paper contribution is not clear.
Our response: We would like to thank the Reviewer for this comment as it highlighted an aspect of this research that, probably, was not clearly stated in the paper. Our objective in this research was to develop a prognostic model that could factor in machine-workload information and compare its performance to a model that does not account for machine-workload. To achieve this, we took advantage of a well-known technique, i.e., logistic regression - for degradation assessment, and the Artificial Neural Network, in a FeedForward configuration, for the prognostics part. The purpose of This comparison was just intended to show which was the performance gain when integrating workload information with respect to a model which do not so. as is currently done. To the best of Authors’ knowledge, none of extant models accounts for workload information.
It is true that this comparison can be extended in the future to include other existing techniques for prognostics. However, the added value of this paper lies in the fact that we have proposed and evaluated the increase in the accuracy of prediction in a situation in which where the workload variability plays a central role. We apologize for have not been clear in our previous answer to the reviewer and appreciate the feedback provided. We are committed to making the necessary revisions to ensure the clarity and comprehensiveness of our work, as reflected in a new revision of the final part of Section 1 ‘Introduction' and the last paragraphs of Section 5 ‘Conclusion’ (the modified parts are now highlighted in red in the new version of the appended manuscript).
Author Response File: 
Author Response.docx
Reviewer 3 Report
It is requested to kindly share the manuscript in which required changes were done and highlighted.
In submitted revised version, it is unclear to notice the changes made by reviewer based on reviewers suggestions.
Author Response
Comments No 1: It is requested to kindly share the manuscript in which required changes were done and highlighted. In submitted revised version, it is unclear to notice the changes made by reviewer based on reviewers suggestions.
Our response: Thanks for this comment. We are sorry for this, and we have submitted a new version of the paper in which the changes are now highlighted in red. We apologize for any inconvenience this may have caused and appreciate the reviewer’s attention to this matter
Author Response File: Author Response.docx
