A New Fault Diagnosis Method for Unbalanced Data Based on 1DCNN and L2-SVM

Round 1

Reviewer 1 Report

Authors conducted this research in the title of “A New Fault Diagnosis Method for Unbalanced Data Based on 1DCNN and L2-SVM".

The paper’s subject could be interesting for readers of journal. Therefore, I recommend this paper for publication in this journal but before that, I have a few comments on the text that should be addressed before publication:

 

Comments:

1)Figure 4, Page 10: Related pictures to "Signal acquisition and processing" and "Verification of the proposed method" are not easily legible and it is hard to read the details. Authors should import larger and more legible pictures. The lack of legibility would cause misunderstanding for the readers.

2)Figure 7, Page 12: With there is enough space in this page, It is better to move these three pictures to the right side and put them in the mid alignment. It really looks better.

3)Abstract: In the Abstract section, authors should pay more attention to the main goals and questions that are supposed to be addressed in this article. It would be really helpful for readers of this paper because they first read abstracts to know if an article interests them or is related to a subject important to them. Instead of checking numerous written materials, readers depend on abstracts to quickly determine if an article is relevant to them or not.

4)Which software has been used in this work to export the charts and diagrams in this work? For instance, software like SigmaPlot or SmartDraw are used to export and depict charts. Mentioning used software would be helpful to future researches and studies in the field of this article.

5) Figure 10, Page 16: Authors used "Shaft" and "Acceleration sensor" words in this figure, but there are no obvious explanations for them. It is better to explain their definitions after or before Figure 10. It is really helpful for the readers of this Article.

6)Page 5: The space between Line 177 and Line 178 is too much compared to other lines. Authors should reduce this gap. Line 177 and Line 178 should be closer than what it is now.

7)Which software is used in this article to model and analyze data? Moreover, which software and indices are used to compare proposed model results with other existing models? Mentioning the used software would be really useful for future works.

For instance, MATLAB and Python are highly utilized by the users to model and analyze data. OfCourse there is extensive range of similar ones and it is optional to use them.

8)Since recently it has been proved that artificial intelligence (AI) and machine learning has a numerous applications in all of engineering fields, I highly recommend the authors to add some references in this manuscript in this regard. It would be useful for the readers of journal to get familiar with the application of AI in other engineering fields. I recommend the others to add all the following references, which are the newest references in this field

[1] Mousavi, N. S., Vaferi, B., & Romero-Martinez, A. (2021). Prediction of surface tension of various aqueous amine solutions using the UNIFAC model and artificial neural networks. Industrial & Engineering Chemistry Research, 60(28), 10354-10364.

[2] Roshani, M., et al. 2020. Application of GMDH neural network technique to improve measuring precision of a simplified photon attenuation based two-phase flowmeter. Flow Measurement and Instrumentation, 75, p.101804.

[3] Zhou, Z., Davoudi, E., & Vaferi, B. (2021). Monitoring the effect of surface functionalization on the CO2 capture by graphene oxide/methyl diethanolamine nanofluids. Journal of Environmental Chemical Engineering, 9(5), 106202.

[4] Vaferi, B., & Eslamloueyan, R. (2015). Hydrocarbon reservoirs characterization by co-interpretation of pressure and flow rate data of the multi-rate well testing. Journal of Petroleum Science and Engineering, 135, 59-72.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

This paper presents a deep learning method using a 1D CNN replacing the ReLU activation function with a smoother SELU and an L2-SVM loss to replace cross-entropy loss to deal with unbalanced data input. 

Their proposition experimentation is based on Industrial motors anomaly detection in driver bearings. It is also supported by cross-comparison experiments using different CNNs, where their proposition thrives in terms of detection accuracy. 

My decision is to accept it as is. 

Very Minor Comments only

There is a - symbol that splits words all over your manuscript. Please also check for syntax and typo errors:

Abstract

Taking aim -> Focusing  

Introduction

nuclear power plant --> plants

impact alternating load (of)

Introduction par. 2 In summary, ..... Please rewrite

Figure 1 probably not needed.

Section 3.4 -> Step 3. The CNN model.. Please rewrite

Section 4 This section uses the Case Western Reserve University (CWRU) bearing datasets --> Please provide reference

Colcusions section. Big single first paragraph. Please amend.

Conclusions section Please amend high recognition accuracy.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

all the comments have been addressed correctly