Estimation and Mitigation of Unknown Airplane Installation Effects on GPA Diagnostics

Round 1

Reviewer 1 Report

This paper presents a framework for a physics-based diagnostic routine with unknown pressure recovery, bleed flow and shaft power extraction. The proposed method has been evaluated for a low bypass turbofan with two different sensor and two different diagnostic schemes. The result seems to validate its effectiveness. Generally, the proposed method provides a potential tool to handle the known/unknown installation effects in real application. Some concerns are listed as follows. 1. The latest researches in the field of Gas turbine diagnostics should be comprehensive reviewed and discussed. From the references, it can be seen that most of the literatures are too old to reflect the recent development. 2. In part 2.4, it is claimed “There are two differences from the published method. The first is that no health parameters are discarded due to correlation to other health parameters. This does not impact the numerical robustness, even though it may be hard to distinguish two or more highly correlated health parameter in the diagnostic evaluation. The second difference is that for each health parameter discarded due to insufficient number of measurements, …”. It is not so clear what the “published method” refers to? What’s more, in the first point, no health parameters are discarded, while each health parameter is discarded in the second point? 3. The selection criterions of some key hyper-parameters are not provided, such as the variables α1, α2, α3 and α4; the input and hidden layer nodes in NN. 4. In the manuscript, the organization of part 2 Methodology can be improved to enhance the logicality and readability. The specific estimation and diagnosis procedure are not well presented. A system summarization of the whole processing procedure is necessary, where a detailed flowchart can be added. 5. It seems that the proposed framework just combines existing techniques to handle the Gas turbine diagnostics. The novelty is not well highlighted. What’s more, the superiority of the proposed method is not fully demonstrated by comparing with existing methods.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The paper presents a model-based approach that uses physical interpretations for gas path analysis in gas turbine engine. An artificial neural network has been involved for the estimation of installation effects and physics-based diagnostics to reduce uncertainties imposed by unknown installation effects. A global reading of the paper makes me say that the article is well written and that the followed methodology is well organized, the English is also very clear. I only recommend these minor corrections:

1) The abstract seems too long, please try to shorten it and make it a single paragraph, to make it simple and easy to understand. Please try to move the detailed explanation to the introduction and also follow the MDPI template for maximum allowed words.

2) From line 73….81, please add more details on limits of previous works, before moving to you contributions and proposed solutions.

3) In the last paragraph of the introduction section, it is best to clarify your contributions and present them as a list.

4) It is recommended to add a paragraph at the end of the introduction section describing the organization of the manuscript.

5) At the beginning of section 2, please insert a small paragraph describing the main points it introduces. Also, please add a flow diagram in a sort of graphical abstract explaining the followed methodology.

6) In table 4, all health parameters have no units, it is recommended to remove the units column.

7) Please follow the MDPI template carefully when inserting figures and tables captions.

8) In section 2.5, it is recommended to show some data visualization concerning the degradation measurements. In this context, please add figures explaining the recorded degradation path in this experiment under the used parameters.

9) Figure 3 and Figure 6 look a bit dry, they lack an artist touch. Please try to improve them by adding more details, and sub-processes. Also, please add colors.

10) In the beginning of section 3 and similar to section 2, please insert a small paragraph describing the main points it introduces.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

The diagnostic method uses a physical model to increase the knowledge in simulation by estimating unmeasured data from a simulator. The model of normal functioning is increased with knowledge of degradation states after solving Gas Path Analysis (GPA) problems. Health parameters, representing degradations, are added to the state vector and relevant measurements are added to the target vector.

Then, the authors use un feed-forward neural network (NN) for estimating bleed flow and shaft power only through available measurements and known operating conditions. The diagnosis is made by comparing the estimates with known reference values.

The results obtained show the effectiveness of the proposed approach

The paper can be improved on the following points:

- The abstract is too long, in my opinion it needs to be reduced to properly locate the contribution of the paper.

- The main contribution of the paper lies in the use of surrogacy, but this approach is not presented in the paper. To give more relevance to this paper, the authors must explain in detail the GPA and how it is used to augment the physical model with degradation states.

- The introduction of paper can be improved by adding diagnostic review papers such as: A survey of fault diagnosis and fault-tolerant techniques; part i: fault diagnosis with model-based and signal-based approaches, IEEE Trans. Ind. Electron. 62 (6) (2015) 3757e3767.
A survey of fault diagnosis and fault-tolerant techniques; part ii: fault diagnosis with knowledge-based and hybrid/active approaches, IEEE Trans. Ind. Electron. 62 (6) (2015) 3768e3774.  As well as papers that use a data-driven model to estimate normal functioning and compare it to measured functioning to generate health indices, such as: Fault detection and isolation in marine diesel engines: A generic methodology. IFAC Proceedings Volumes 45 (20), 964-969.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

All my issues have been addressed.

Reviewer 3 Report

The authors did not answer all my questions, which shows their expertise in the subject.