Analysis and Prediction Research for Bipropellant Thruster Mixture Ratio Based on BP-RNN Chain Method

Round 1

Reviewer 1 Report

1. Main concern of this reviewer on this article is that the research seems to be based on not clear or not physically correct data. For example, 480 thrusters classification does not seem logical without an understanding on the specification of each thruster.

2. There are a few abbreviations that were used before they were defined. The authors should check the manuscript thoroughly from this point of view.

3. Would it be appropriate to assume that potential readers would understand what ReLU function, Adama optimization, LeakyReLU function, CrossEntrpyLoss function, Softmas layer, Adagrad optimizer, etc? If not, before they are used, there should be at least a brief introduction on them and why the authors considered them to use in the study. 

4. Line 28, "Bipropellant propulsion system has been widely used in a large number of high-orbit satellites for attitude and orbit control.."

This statement requires evidence with proper citations.

5. Line 30, "During the orbital transition, constant pressure mode is commonly used to supply propellant for apogee engine, and propulsion system can be transferred to dropping pressure mode for thrusters with operation of attitude control and position maintenance after into orbit."

This statement also requires also support particularly on the 'Dropping pressure mode' 

6. Line 33, "This process design can not only provide adequate total impulse to orbit, but also improve control accuracy and working reliability in orbit [1-4]."

It does not seem logical to say 'dropping pressure mode' will contribute to 'adequate total impulse'. Additional information should be added to support and clarify the statement.

7. Line 45, "The propellants widely used in the bipropellant propulsion system of spacecrafts are NTO (nitrogen tetroxide) as oxidizer and MMH (methyl hydrazine) as fuel."

Not all potential readers would agree on this statement, the authors should add more information on this.

8. Line 46, "The optimal mixture ratio of NTO and MMH is 1.65, which is also the ratio of the density of oxidizer to fuel"

It is unclear how the optimum mixture ratio is also the density ratio of oxidizer to fuel.

9. Line 61, "It shows that the mixture ratio accuracy requirement of bipropellant thruster is quite high"

(A) First, it is unclear why the mixture ratio range is required as shown in the table 1. The authors should double check if it was from a reliable source, and why those mixture ratios were required for what reason with more details.

(B) It is unreasonable to consider it 'high' just because of one reference. More literature surveys would make it look more reliable.

10. Line 70, "There is no logical correlation between these two kinds of tests, which makes it difficult to learn the association rules and master the adjustment law, as shown in Figure 1. Therefore, more than 60% of thrusters need to correct the orifice size after firing test, but those results after correction cannot be verified by firing test any more."

(A) It is unclear where figure 1 came from. Proper citation should be added.

(B) It should be clarified on what experimental condition the data were gathered.

(C) Unclear why more than 60% of thrusters should be corrected.

(D) It is also unclear why there are those distributions with that particular mixture accuracy without any details described.

11. Line 80, "In the liquid-flow test, alcohol is used to be test medium."

Does this statement mean alcohol is generally used for bipropellant liquid flow tests?

12. Line 100, ".. BP-RNN chain .."

There should be an introduction to its meaning and relevant previous studies before the abbreviation 'BP-RNN' was used.

13. 2.1. Dataset construction. ".. 10-dimensional test results were selected.."

(A) What are the 10-dimensional test results and where they were from?

(B) There is a weak logical flow in 2.1. section due to 'Injector welding' that appeared out of sudden. There should be clarification on the factors that affect the mixture ratio first, and then additional description on the 'injector welding' as well.

14. It's hard to understand figure 2 as well. More information on this data is required: where it came from, in what condition it was acquired, what were input variables considered that affected the data, and what is the meaning of it.

15. Line 139, "The hidden  layer adopts a three-layer structure with 3000, 2000 and 1000 neurons from the first to the 140 third layer."

Was there any particular reason that they were 3000, 2000, and 1000? any random values would be also acceptable?

16. Line 166,  "As seen in Figure 3(a), for all enhanced data, 60.41% of the fitting results are completely accurate, and 37.39% have small classification deviation of only 1 which was also acceptable."

With what criteria was it evaluated as 'completely accurate'?

 

17. Line 187, "This reflects that injector welding process has a great effect  on the NTO flow channel."

(A) A potential reader might be curious that if the statement is true, this might have been possible to be said with just the raw data even before applying BP RNN chain method. (B) What is the main reason the method was employed? (C) with what novelty?

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

The authors have conducted research and made some interesting points related to the Bipropellant Thruster Mixture Ratio. I think the following issues can be explored. 

1. It is suggested that the authors add a chart comparing the strengths and weaknesses of the existing methods in the introduction to show the characteristics of the existing research concisely and visually.

2. The authors have used the BP neural network algorithm. However, BP neural networks have been around for a long time. Have the authors tried any of the latest AI algorithms available?

3. In the Data and Results section, it is suggested that the authors add a comparison of critical parameters between the new method proposed in this paper and existing research. The advantages of the new method over existing methods should be visually illustrated.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

 

The work deals with very important issues related to the combustion process. A very innovative research method was used. The work is at a very high level. The authors say that in order to improve the accuracy of the blend ratio of the dual-fuel thruster and extend the life of the satellite in orbit, a variety of liquid flow tests and a final burnout test are critical in the production testing process. The BP-RNN chain method based on machine learning was used, which uses multivariate non-linear parameters to build a specific data set after data improvement. The tests show the high effectiveness of the research method used. The introduction is prepared correctly. A detailed analysis of the topic was performed. Many complex analyzes have been carried out, and many works by other authors have been cited. The method of solving the problem was correctly selected. The system of data entry and achieved results is presented in detail. The drawings have been properly prepared. The results and discussion. The authors undertake very detailed research. The achieved results, according to the attached drawings and charts, indicate that the model has been made well. Figure 6 shows the comparison of the effect of the normalized weight of each factor. The summary aptly includes the most important conclusions. The percentage values ​​of the achieved accuracy for selected cases are given. I do not see any factual errors. Good quality work.You can increase the number of citations

Author Response

Thanks very much for reviewer’s high evaluation. According to reviewer’s suggestion, more citations are added to improve article’s quality.

Round 2

Reviewer 1 Report

Comments and questions raised in the previous review process have not been properly addressed yet. For example, the main data source has no clear reference. If it was tested by the authors, then clear experimental test conditions must be provided.

There is still a weak logical flow. Out of a sudden, a welding issue appears, and it is totally not logical to point out the issue from the data without any experimental test details. It can be right or not right as nothing support the conclusion.

Some answers to the previous comments/questions are irrelevant and hard to understand what was meant. The authors should do some more literature surveys on rocket propulsion and use a proper word that is commonly used in the rocket field. For example,  'dropping pressure mode' is not what people use for a rocket study.

Having some more literature survey will also be helpful for a better manuscript, as there are some statements with only a few references or no references. For example, the statement, "The combination of MMH and NTO is currently the most widely used propellant selection", is not supported by any evidence.

With a few statements that are irrelevant, it is clear that the manuscript has been written without sufficient literature survey and background on rocket propulsion before the BP RNN was applied.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

No more comments.

Author Response

Thanks a lot for reviewer's help. 

Round 3

Reviewer 1 Report

The revised manuscript with more details looks better than the previous version. There is however still unclarity, for example, on the 480 thrusters, if they have any common test condition, what were the thrusters for what purposes, and with what design details, so that the authors can analysis anything on the thrusters. There must have been clear control variables and fixed variables for the 480 thrusters' design parameters and operating conditions so that meaningful results can be made. The same is true for the injector, to discuss about injector welding. The manuscript would look better with more information and clarity considering these points mentioned and it might be in quality for publication after that.

Author Response

Please see the attachment.

Author Response File: Author Response.docx