Diesel Engine Fault Diagnosis Method Based on Optimized VMD and Improved CNN

Round 1

Reviewer 1 Report

This manuscript presents an approach using a novel method using optimal variational mode decomposition (VMD), (CWT), and improved convolutional neural networks (CNN) to identify faults in diesel engines using processed vibration signals.

The potential of the proposed method in fault diagnosis tasks of diesel engines is tested using experimental signals operating at the same running conditions. The results are then compared to the results with other well-known methods from the literature.

The authors clearly addressed the problem based on the fair and updated review of literature they conducted. The novelty of work is good enough, however, there are some issues that need to be revised by the authors to improve the contribution and the quality of this manuscript as follows:

1. In the abstract section (line 11), the authors wrote “ideal variational modal decomposition (VMD).” What do you mean by ideal? Also, it is referred to as “mode,” not “modal.”

2.  In the introduction part (line 31), the authors wrote, “uncooked vibration signal…”. What do you mean by that? Do you mean raw vibration signal? Please clarify by using more technical terms. The same applies in line 29 for choosing the word “paralysis” and in line 40, “mannequin.”

3. In line 93, the authors should correct “variable modal decomposition” into “variable mode decomposition.” The lines 94 and 95 should be rewritten and for clarification.

4.  The authors should add more papers in the review of the literature concerning the efforts made to optimize VMD algorithm and adaptive/blind selection of its parameters.

5. The authors mentioned in the title, introduction section and in conclusion part that they used optimized VMD. They used scattering entropy to select the optimum number of decomposition layers. However, a preliminary step in applying the VMD algorithm is to optimize the number of modes (k) and the data quadratic parameter (α) value for each raw input signal. These VMD parameters control the instantaneous frequency bandwidth of the mode, which reduces the chances of mode mixing problems and inhibits the presence of background noise in the relevant modes. To claim VMD algorithm as optimized, these parameters should be calculated for each raw signal. If not, then we can’t call it an optimized algorithm. Could you please refer to that more clearly in section 2.1?

6.     Figure 5 needs to be improved for readability and clarity.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

This paper shows the diesel engine fault diagnosis technique with optimized VMD and 469 improved CNN. The research work looks interesting. The manuscript can be accepted for publication after addressing following suggestions:
Comments:
1)Improve the quality of figure 8.
2) Include the engine specification in the experimental section..

3) Engine test procedure can be explained in detail in the experimental section

 

4) Kindly include the Uncertainty analysis for the experiments.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf