A Review of Statistical-Based Fault Detection and Diagnosis with Probabilistic Models

Round 1

Reviewer 1 Report

There are some major concerns about the paper:

1-      First of all, it is proposed to consider a comprehensive review on the English writing of the paper.

2-      My second major concern is on the structure of the material. The paper is written as a collection of abstracts, with a very generic introduction and a discussion built around a summary table with very little information in it. Long summaries are included for each paper and the result is a very long paper with little discussion/justification for the methods presented or comparison between approaches. This is the part that needs substantially more work. The long summaries should be reduced in the length, and the focus should be on the big picture. There is a long paragraph per paper on sections that go on for many pages, and little or no cross-comparison of the schemes. This needs quite substantial rewriting. There is very limited contextualization of the papers reviewed, and almost no critical analysis. I also propose review current papers like model-based fault detection and identification of a quadrotor with rotor fault.

3-      Following on this, İt needs a clearer definition of the fault detectıon problem.

4-      There is minimal discussion on sensors and thir requirements.

5-      Regarding model-based methods, see the recent paper :10.1007/s42405-023-00633-0.

6-      fault detection strategies is not very clear and features some repeated sentences. Moreover, in my opinion, it is not necessary to give the short explanation of different (well known) fault detection techniques. Fianlly, I am not convinced this classification is the most useful/clear.

7-      The discussion on the current limits and trends is quite limited and shallow. In my opinion, it can be improved to give a better insight into the gaps and why they are there. The fact that few works adopt a specific control technique does not automatically imply that there is a research gap there.

8-      In some cases, there are wordy descriptions (not always clear) that can be replaced by equations or pics.

9-      It's essential to include a comparative analysis of these statistical techniques, discussing their performance metrics, applicability under different scenarios, and potential limitations. This comparison will aid readers in assessing the relative effectiveness of each method. Using tables and important comparison criteria is highly recommended.

10-   The inclusion of discussions on current challenges and future trends in FDD is commendable. However, the abstract should briefly highlight a few specific challenges faced in implementing these probabilistic models in real-world industrial settings. Additionally, providing insights into anticipated advancements or research directions would enhance the paper's relevance.

11-   It would be beneficial to include examples or case studies showcasing practical applications of these probabilistic FDD models in various industries. This addition can illustrate how these methods have been utilized, their successes, and potential hurdles encountered in real-world implementations.

No comment

Author Response

Dear reviewer, please see the detailed response in attached pdf file.

Author Response File: Author Response.pdf

Reviewer 2 Report

The manuscript titled “A Review of Statistical-based Fault Detection and Diagnosis with Probabilistic Models” presents a comprehensive review of probabilistic fault diagnosis. for use in industrial applications. The purpose of this review is to provide a theoretical background and recent applications of probabilistic statistical fault diagnosis.

In the present draft of the manuscript, the authors conducted a comprehensive review of existing scientific literature. While I do not deny the relevance or value of these studies, I believe that it would be beneficial for the paper under review if the authors were to clarify the criteria for selecting the referenced papers. Providing this information would enhance the overall quality of this paper. Furthermore, I suggest that the authors consider incorporating more recent research to reflect on the advancements in technology that have occurred in recent years.

An area that was conspicuously absent from the authors' survey is the field of probabilistic models related to multichannel profile data from multiple sensors during a fault event. I recommend that the authors include this important domain in their study.

To enhance the overall quality of the manuscript, I suggest that the authors conduct a literature review that emphasizes the contribution, approach, and key results of each cited scientific work along with a brief description of the existing unresolved issues in the referenced studies.

Because the manuscript contains equations that were not originally developed by the authors, I recommend that they be properly cited. Additionally, for any equations developed by the authors for the first time, it is important to provide a detailed explanation to ensure that readers can fully understand the reasoning behind them.

It will benefit the manuscript if the authors devise a proper "Conclusions" section in which they state the most important outcome of their work. The authors should not restate what they have done or what the article does; instead, they should focus on what they have discovered and most importantly on what their findings mean.

Author Response

Dear reviewer, please see the detailed response in attached pdf file.

Author Response File: Author Response.pdf

Reviewer 3 Report

1. Before starting Section 1.1, it's necessary to have a brief discussion on the problem statement.

2. In Figure 1, what is the difference between the disturbance and noises? we could categorize noise as a disturbance. 

3. Figure 2 requires more details and discussions. 

4. In section 3, several models have been reviewed. Thus, a table for comparison of them is essential. Moreover, several practical examples with details could be useful. 

5. What is the purpose of section 6? Several lines on it should be written before section 6.1.

6. The conclusion is general and can be more specific for readers who want to start researching this area. 

Author Response

please see the PDF

Author Response File: Author Response.pdf

Reviewer 4 Report

1. In section 1.3, motivation and contribution, explain why this review is needed in detail.  since there are a lot of review papers in fault detection and diagnosis, explain what your review paper consists of, which are not covered by those articles. Also, cite the popular review papers in fault detection and diagnosis, like: 'A review on fault detection and diagnosis techniques: basics and beyond', 'Bearing Fault Detection and Diagnosis Using Case Western Reserve University Dataset With Deep Learning Approaches: A Review', 'Challenges and opportunities of deep learning-based process fault detection and diagnosis: a review', 'A review of computing-based automated fault detection and diagnosis of heating, ventilation and air conditioning systems', and so on. Mention what they lack, and how you are different from the existing review articles. 

2. The majority of the article is on the theory of the different models mentioned. I believe these are not developed by the authors. also, there is a huge literature on the theory behind these models. The main focus should have been on overviewing the articles and reviewing them critically. For the reviewing part, one or two sentences are used per article. I do not think this approach will provide readers the enough knowledge.

3. Section 5 says 'Challenges and Recent Applications on Statistical Fault Diagnosis'. I guess that should be Recent applications and then you should discuss the challenges. That will be systematic. 

Better

Author Response

Please see the PDF

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

The authors of this manuscript have made revisions to address the concerns identified in the previous iteration.

Author Response

Author Response File: Author Response.pdf

Reviewer 3 Report

I don't have any further comments. 

Author Response

Thank you very much  pls see the pdf

Author Response File: Author Response.pdf

Reviewer 4 Report

'5.3. Approaches for Non-stationary' non-stationary what?

Moderate changes

Author Response

Thank you very much pls see the PDF

Author Response File: Author Response.pdf