Next Article in Journal
Streamlined Bearing Fault Detection Using Artificial Intelligence in Permanent Magnet Synchronous Motors
Previous Article in Journal
Investigation of an Optimized Linear Regression Model with Nonlinear Error Compensation for Tool Wear Prediction
 
 
Article
Peer-Review Record

Simulated Centrifugal Fan Blade Fault Diagnosis Based on Modulational Depthwise Convolution–One-Dimensional Convolution Neural Network (MDC-1DCNN) Model

Machines 2025, 13(5), 356; https://doi.org/10.3390/machines13050356
by Zhaohui Ren *, Yulin Liu, Tianzhuang Yu, Shihua Zhou, Yongchao Zhang and Zeyu Jiang
Reviewer 1: Anonymous
Reviewer 2: Anonymous
Machines 2025, 13(5), 356; https://doi.org/10.3390/machines13050356
Submission received: 31 March 2025 / Revised: 21 April 2025 / Accepted: 23 April 2025 / Published: 24 April 2025
(This article belongs to the Section Machines Testing and Maintenance)

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

1) In Figure 1, it cann't found the relationship of DWconv, CMM and MDC-1DCNN. It's difficult to understand the logic of the proposed method.
2) As shown in Table 4, there are 3 fault types in the experimental datasets, which is a bit small. Why not provide more fault types (like the simulation) in the experimental dataset ? If it can study above 5 fault types in the experimental dataset, the simulation can be avoided to conduct.
3) Some writing errors in the text need to be corrected, such as, CMM should display its full name when it first appears in the main text.
4) The paper studied the fault diagnosis of Centrifugal fan blades. Can it also be effective for other components of centrifugal fans?
5) Some latest literature on fault diagnosis of rotating machinery can be cited, such as, 10.1504/IJHM.2024.135976, 10.1016/j.ress.2025.110847.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

This paper proposes a novel method for centrifugal fan blade fault diagnosis based on a modulational depthwise convolutional neural network (MDC-1DCNN). The model combines local and global signal feature extraction using a convolutional modulation module and multi-scale depthwise convolutions. Experimental results on both simulated and real-world datasets demonstrate superior performance compared to existing methods. The topic is highly interesting, particularly from the perspective of fluid machinery integration with AI techniques. The paper is of good quality and well-structured, though some improvements are needed.

The introduction is a quite complete. The introduction is quite comprehensive, providing a brief overview of the state of the art on centrifugal fans and, most importantly, on convolutional neural networks. However, this section it could be further strengthened by incorporating additional references to CFD studies, particularly those related to turbomachinery applications across various domains. Moreover, it would be valuable to include citations on uncertainty quantification (UQ) methodologies. Given the stated objective of developing a digital twin, it would also be beneficial to reference relevant examples of machine learning (ML) applications in CFD simulations—such as those used for surge detection. For example, you might consider citing:

  • Carrattieri, L.; Cravero, C.; Marsano, D.; Valenti, E.; Sishtla, V.; Halbe, C. “The Development of Machine Learning Models for Radial Compressor Monitoring With Instability Detection". Journal of Turbomachinery, 2025, Vol. 147, Issue 5, p. 051004.
  • Hammond, J., Pepper, N., Montomoli, F., & Michelassi, V. (2022). Machine learning methods in CFD for turbomachinery: A review. International Journal of Turbomachinery, Propulsion and Power, 7(2), 16.

The methodology is thoroughly described, with a clear focus on the details of DWconv, CMM, and the MDC-1DCNN framework. However, the schematic diagrams presented are somewhat difficult to interpret and would benefit from improved clarity.

In the experimental setup section, additional details on the instrumentation used—along with their measurement tolerances—should be provided. Moreover, the geometrical information of the experimental setup should be clearly included, preferably within the figures.

The results section demonstrates the high accuracy of the proposed methodology. Nevertheless, the discussion of these results could be expanded with deeper insights and more comprehensive analysis.

The conclusions effectively highlight the main contributions of your work.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Authors have improved the manuscript according to the reviewers' comments, and the quality and the quality and readability have been significantly enhanced. Therefore, I believe it can be published.

Reviewer 2 Report

Comments and Suggestions for Authors

All my comments and suggestions have been correctly addressed and added in the revised work. Now it is ready for the publication in this form by havenig increased its quality.

 

Back to TopTop