Research on a Noise Reduction Method Based on Multi-Resolution Singular Value Decomposition
Round 1
Reviewer 1 Report
The presented method is good and is well writtend.
I suggest that the experimentla work is required to validate your method.
Authors shoud find any application such as gear fault or other industrial engineer field.
Author Response
Dear Editors and Reviewers:
Thank you for your letter and for the reviewers’ comments concerning our manuscript entitled “Research on Noise Reduction Method Based on Multi-Resolution Singular Value Decomposition”(ID: 695772). Those comments are all valuable and very helpful for revising and improving our paper, as well as the important guiding significance to our research. We have studied comments carefully and have made correction which we hope meet with approval. Revised portion are marked in red in paper. The modified manuscript has been uploaded. The main corrections in paper and the responds to the reviewer’s comments are as flowing:
Response to Reviewer 1 Comments
Point 1: The presented method is good and is well written. I suggest that the experimental work is required to validate your method. Authors should find any application such as gear fault or other industrial engineer field.
Response 1: In order to make the paper more convincing, we have added the faulty bearing experiments of Case Western Reserve University to verify the method in this paper. We decomposed the signals of the outer ring of the rolling bearing by MRSVD to obtain 5 layers of detailed signals. We selected the detailed signal with the largest kurtosis value for envelope spectrum analysis, and compared it with the other 5 noise reduction methods. It was found that the fault signal processed by MRSVD reduced the noise more effectively, and the fault frequency of the bearing was better extracted, which verified the method in this paper.
Reviewer 2 Report
This paper presents noise pollution reduction based on a multi-resolution singular value decomposition method. The following comments need to be addressed before acceptance.
Paper was not formatted according to the journal guidelines. The paper is too long. The theoretical section could be shortened significantly. The flow chart can be taken to the appendix. That might help a reader to go over the paper smoothly. In the abstract, a summary of the results needs to be added. The conclusion section is too big, please make it short and precise. How the experimental signals are obtained? If it simulated experimental signals with added noise (Fig. 3) need to be described clearly. The results need to be presented in a more organized way. It is difficult to follow the results and discussion. Reducing some of the results might help.Author Response
Dear Editors and Reviewers:
Thank you for your letter and for the reviewers’ comments concerning our manuscript entitled “Research on Noise Reduction Method Based on Multi-Resolution Singular Value Decomposition”(ID: 695772). Those comments are all valuable and very helpful for revising and improving our paper, as well as the important guiding significance to our research. We have studied comments carefully and have made correction which we hope meet with approval. Revised portion are marked in red in paper. The modified manuscript has been uploaded. The main corrections in paper and the responds to the reviewer’s comments are as flowing:
Response to Reviewer 2 Comments
Point 1: Paper was not formatted according to the journal guidelines.
Response 1: I am very sorry for not formatting in the correct format, now we have reformatted the paper in the format required by the journal guidelines.
Point 2:The paper is too long. The theoretical section could be shortened significantly.
Response 2: After discussion, we have made a lot of modifications to the theoretical part. We have partially deleted the last paragraph of the introduction, deleted the theoretical part of the SVD algorithm in section 2.1 of the old manuscript, and deleted the 121-125 lines of the two-division recursive SVD, the 176-184 lines of multi-division structure MRSVD. In addition, at the end of section 3.1, the factors affecting the noise reduction rate were discussed in the old manuscript. Because this part is not closely related to the title of section 3.1, the old manuscript lines 347-366 are deleted.
Point 3: The flow chart can be taken to the appendix. That might help a reader to go over the paper smoothly.
Response 3: The flow chart of the figure 3 has been attached as an appendix at the end of the paper.
Point 4: In the abstract, a summary of the results needs to be added.
Response 4: We modify the abstract and add a summary of the whole paper's work to it. Firstly, we analyze the noise reduction of multi-division structure MRSVD, and get the optimal noise reduction model. Then, we apply the optimal model to the fault bearing diagnosis. Compared with other methods, we verify that the selected optimal model has better noise reduction effect and can effectively diagnose the fault bearing.
Point 5: The conclusion section is too big, please make it short and precise.
Response 5: In order to make the conclusion more clear and specific, we refine the core content of the article, and at the end of the first paragraph of the conclusion, we summarize the work and contribution of this paper.
Point 6: How the experimental signals are obtained? If it simulated experimental signals with added noise (Fig. 3) need to be described clearly.
Response 6:
The two experimental signals are as follows:
Experimental signal 1: x1=sin3t+sin20t, the sampling time is 2π, sampling frequency is 1024HZ, the number of sampling point is 1024, the frequency of the signal is f1=0.477 and f2=3.183 respectively.
Experimental signal 2: x2=sin(2πt)+sin(20πt)+cos(50πt)+sin(100πt), the sampling time is 1, the sampling frequency is 1000HZ, the number of sampling point is 1000, the frequency of the signal is f1=1, f2=10, f3=25, f4=50 respectively. Then we add white noise with SNR of 5dB, 1dB, 5dB, 20dB and obeying the Gaussian distribution N(0,1) respectively, and get four noisy signals, which are noise1, noise2, noise3, noise4 in the old manuscript flow chart (figure 3). These four noisy signals are used to select the optimal model of the simulation signals.
Point 7: The results need to be presented in a more organized way. It is difficult to follow the results and discussion. Reducing some of the results might help.
Response 7: After discussion, we delete the part of the discussion on the factors affecting the noise reduction rate, and add the application example of rolling bearing fault diagnosis after Section 3.2 of the old manuscript, which makes the content of the paper more substantial. Finally, the results and discussion are modified to make it clear and more rational.
Round 2
Reviewer 2 Report
Paper can be accepted for publication.
