Hydroelectric Unit Vibration Signal Feature Extraction Based on IMF Energy Moment and SDAE

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Thanks for the contribution. The paper deals with an interesting topic about a new approach to extraction to feature of hydroelectric unit vibration signal. The main idea of developed approach is used the IMF energy moment to the diagnosis of rotating machinery and the stacked denoising auto encoder (SDAE) as an unsupervised feature-learning model. The proposed approach consists of three main steps: 1) the typical time-domain and frequency-domain characteristics of vibration signals are calculated; 2) the time-frequency characteristics of vibration signals are used; 3) the stacked denoising auto encoder is used.

There are some concerns related to the manuscript:

The proposed approach is quite difficult. It should be noted in a separate paragraph how applicable the proposed solutions are in practice.

Specific comments:

1.      Rules for tuning of the proposed method should be given. How these changes will be affect the obtained results. This can be explained on the basis of Fig. 5.

2.      Line number 73: At present, there are various types of hydropower units. Its have significant differences in static and dynamic capabilities. In what way it is necessary to change the proposed method when changing the investigated unit.

3.      Line numbers 63-75: The advantage of the proposed approach over existing solutions needs to be clearly defined.

4.      Line numbers 186-212: How the reliability of the information obtained by the developed method is verified.

Author Response

Comments 1: Rules for tuning of the proposed method should be given. How these changes will be affect the obtained results. This can be explained on the basis of Fig. 5.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The study "Hydroelectric unit vibration signal feature extraction based on IMF energy moment and SDAE" addresses an interesting research problem. The authors focused their attention on the problem of assessing the performance of hudroelectric power plants. Noting the difficulty of the subject evaluation by means of a single vibration signal feature under the influence of many factors, the authors proposed an analysis of the performance evaluation in the time-frequency dimension and learning models. For research purposes, the authors applied a multidimensional information fusion feature construction method for hydropower vibration signals based on statistical theory, CEEMD time-frequency analysis method and unsupervised learning model. The proposed method provides a basis for obtaining multidimensional complementarity of feature attributes, which in turn helps to accurately distinguish between types of equipment states, providing a basis for further state identification and trend prediction. This is a very interesting approach worthy of scientific recognition

The abstract of the paper presents the subject of the study and the methodological scope. The abstract signals the gap in the literature to which the paper responds and indicates the novelty dimension of the study. 

The introductory section presents the research background, scope and methodology. It would be worthwhile to expand this section to include a broader background of the problem explored, anchoring the choice of the research subject. The above is justified by the fact that the article does not have a separate section devoted to literature review, and there are few references 19. I believe that this scope should be strengthened.

Item 2 Integrating Multidimensional Information Features for State Recognition of Hydroelectric Units was developed in problem sections. IMF Energy Moment Feature Extraction Based on CEEEMD (Section 2.1), Feature extraction based on stacked denoising autoencoder (Section 2.2) and Hydropower unit state recognition method based on multidimensional information fusion features (Section 2.3) were discussed. Sections 2.1 oracised legibly - no comments. Section 2.2 - line 186 - text needs to be completed under the figure. Section 2.3 needs refinement - lines 194, 198, 204 - the message "Error! Reference source not found". In addition, line 213 - under the figure, the text should be supplemented.

Item 3 Verification and analysis, like section 2.2, needs refinement. It appears numerous references "Error! Reference source not found," which makes it difficult to analyze the text. In addition, under Figure 13 (line 310), the text needs to be completed.

As signaled earlier, the study lacks extended literature work. The discussion section is missing before the conclusion (at the level 4 point of the paper). The authors' results should be analyzed against the background of existing studies available in the literature. This element is supposed to indicate the novelty of the study and its relevance to practice. Similarly, the summary point. It should be more strongly related to the purpose of the work, the gap and the importance of the research. The summary section in this regard should be expanded.

The presented study is interesting and generally well prepared. However, the study should be improved before publication. 

In conclusion, the literature review should be developed, the text should be supplemented in the highlighted places, the summary should be strengthened. In addition, the missing text under the graphics should be completed.

 

 

 

 

Author Response

Comments 1: The introductory section presents the research background, scope and methodology. It would be worthwhile to expand this section to include a broader background of the problem explored, anchoring the choice of the research subject. The above is justified by the fact that the article does not have a separate section devoted to literature review, and there are few references 19. I believe that this scope should be strengthened.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

No more comments. Thanks.