Modeling a Practical Dual-Fuel Gas Turbine Power Generation System Using Dynamic Neural Network and Deep Learning
Round 1
Reviewer 1 Report
The authors of the article presented a work on the subject Modeling a Practical Dual-Fuel Gas Turbine Power Generation System Using Dynamic Neural Network and Deep Learning.
The introduction meets the requirements applicable to scientific articles. The authors reviewed the literature and indicated step by step information discussed later in their work.
The purpose and problem of the work were not found, which makes it difficult to properly assess it.
The authors used real data for the research. The results used for the simulation should be available in typescript.
See line: 67: correct the notation in Figure 3.
See line 174: correct the notation in Figure 4.
See line 266: enter the correct drawing number
See line 300: enter the correct drawing number
See line 401: enter the correct drawing number
See line 417: enter the correct drawing number
See line 429: enter the correct drawing number
See line 476: enter the correct drawing number ... and so on
See line 321: The pattern reference has an invalid number. You have to correct it
See line 629: table number incorrect
Point 5 should be expanded. The presented results have not been analyzed. 6 drawings have been identified that should be properly analyzed.
Conclusions point 6 is too general and presents only selected results. Has the goal of the work been achieved? The authors do not mention this.
Quite a lot of unfinished or "broken" elements, as if from other studies, were found in the work. The typescript should be put in order and corrected.
Author Response
Please see the attachment
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The paper appears well written, and the numerical experiments part is sufficiently rigorous. However, the readability of the manuscript should be strongly improved:
- Many figures have been reported. Reducing the number of figures could improve the readability of the article.
- Many abbreviations have been used. A short dictionary in the appendix would be helpful.
- Some misalignments in the tables are visible. For example, see Tables 1 and 2.
- References to tables, figures and equations are not homogeneous. The authors must solve this issue.
- The description of the gradient descent method and the loss function used should be presented in the same paragraph of the neural networks.
- Line 580: the term "weight sharing function" is distorting. I suggest using the term "mechanism".
- The notation in the different paragraphs appears inconsistent and needs to be improved:
- According to the notation used in Equation 1, Equation 2 should use x instead of X.
- Equations 4 and 10 are not aligned.
- Equation 7 should be rewritten in terms of y and \ hat {y}.
- Equation 14 should have the same notation as Figure 20.
- Line 554 should be (x_1, x_2,…, x_n).
- Line 584 f (.) should be f (\ cdot).
- The caption of Figure 19 should be corrected.
- Please check the numbering of the equations.
Numerous typos are visible. Some examples:
- Line 9 "is" should be "are"
- Line 77 "a single"
-Line 92 should be "a maximum"
-Line 442 "one-neuro" should be corrected.
Punctuation needs improvement:
-Line 361:a dot is missing.
-Line 364, the dot is missing.
Finally, the work investigates the application of some deep learning techniques to the Gas turbine power generation system. Future work's paragraph could insert a reference to other deep learning techniques such as recurrent neural networks and locally-connected neural networks.
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments on Sustainability-1495065 Major comments: The study is addressing “Modeling a Practical Dual-Fuel Gas Turbine Power Generation System Using Dynamic Neural Network and Deep Learning”. The author has studied This paper presents models with extremely accurate simulations for a real dual-fuel gas turbine using different types of Artificial Neural Networks (ANN), deep and dynamic ANNs. The dynamic neural network has been realized via a nonlinear autoregressive network with exogenous inputs (NARX) ANN and the deep one has been based on a Convolutional Neural Network (CNN). The outputs selected for simulations are: the output power, the exhausted temperature, and the turbine speed or system frequency, whereas the inputs are the natural gas (NG) control valve, the pilot gas control valve and the compressor variables. The data-sets have been prepared used in three essential ways for training and validation of the networks: normalized data, standardized data and SI units’ data, in addition to rigorous effort for wide-range trials in tweaking the network structures and hyper-parameters, which eventually lead to highly satisfactory results for both models. The results have shown comparable performance for both dynamic NARX ANN and the CNN with a slight superiority of NARX. It can be then argued that the dynamic ANN is better than deep learning ANN for time-based performance simulation of gas turbines (GTs). However, the results shown in this study are lack of innovation and novelty. The headings also need changes as they should be based on obtained results instead of the equipment used. The English language used in the manuscript needs improvements, as there are some punctuation and grammatical mistakes throughout the manuscript. Sentences need more clarity and better construction. Overall, the manuscript needs some structural and literature revisions to meet the requirements of the journal.
The authors are advised to address the following comments carefully. Some specific comments:
1. Abstract: It is suggested to add some background with a few objectives and possible applications of this study and highlight the novelty of this work. The abstract only contains some parameters without any process conditions or key values from results, which is insufficient to delineate the whole pictures of contribution and possible application of this study. Should add key values from results in the abstract.
2. Please insert the minimum numbers (1-2) after decimal point.
3. In Table-2 there is an error in Frequency units and values.
4. A lot of formatting and grammatical mistakes are present in the manuscript. Carefully, revise it.
5. Revise keywords add more specific and novel keywords with broader meanings (5-7 words).
6. The introduction lacks sufficient background information, which is unable to give the reader detailed background knowledge and possible wide application of this study. Research gaps should be highlighted more clearly and future applications of this study should be added.
7. Add some references in the Introduction section to strengthen the literature review
• Murugesan, A.; Umarani, C.; Subramanian, R.; Nedunchezhian, N. Bio-diesel as an alternative fuel for diesel engines—A review. Renew. Sustain. Energy Rev. 2009, 13, 653–662.
• Teoh YH, How HG, Sher F, Le TD, Nguyen HT, Yaqoob H. Fuel Injection Responses and Particulate Emissions of a CRDI Engine Fueled with Cocos nucifera Biodiesel. Sustainability 2021;13:4930. doi:10.3390/su13094930.
• A. Ayanoglu, R. Yumrutas¸ , Production of gasoline and diesel like fuels from waste tire oil by using catalytic pyrolysis, Energy 103 (2016) 456e468, https://doi.org/10.1016/j.energy.2016.02.155.
• K. Thangavelu S, M. Arthanarisamy, Experimental investigation on engine performance, emission, and combustion characteristics of a DI CI engine using tyre pyrolysis oil and diesel blends doped with nanoparticles, Environ. Prog. Sustain. Energy 39 (2020) 1e7,https://doi.org/10.1002/ep.13321
• Yaqoob H, Teoh YH, Ud Din Z, Sabah NU, Jamil MA, Mujtaba MA, et al. The potential of sustainable biogas production from biomass waste for power generation in Pakistan. Journal of Cleaner Production 2021;307:127250. doi:https://doi.org/10.1016/j.jclepro.2021.127250
8. Please critically discuss your results with the other experimental published data?
9. Most important: Authors should prove their scientific originality by defining what are their main scientific findings, which have not yet been presented in other studies.
10. The tables/figures inserted are not explained or discussed well in the text please discuss critically and explain all tables/figures in the text.
11. Revised the captions of tables and figures. These are very generic.
12. There are some grammatical issues, particularly in the introduction, and results & discussion section.
13. Between the lack of methodology and limited novelty, the authors have a significant amount of work to do
14. The conclusion is only talk about other studies, which is insufficient to depict the whole picture of your contribution of this study. The authors are advised to write the conclusions in a comprehensive way (short) and should contain key values, suitability of the applied method, the major findings, contributions and possible future outcomes (250-300 words).
15. References: The author is being advised to revise this section, there is some missing information. Add some latest references.
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Reviewer 4 Report
The article presents simulation studies with the use of various
neural networks and analyzes their usefulness on the basis of
effectiveness.
The turbine engine is provided only as an example to conduct simulation
tests.
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
The authors addressed my comments.
Reviewer 3 Report
The manuscript has been significantly improved.
