Sensor Fault Tolerant Control for Aircraft Engines Using Sliding Mode Observer
Round 1
Reviewer 1 Report
Review comments:
The paper is well organized with presentation of background information, the mathematical representation of proposed method, an example of practical implementation and lastly, a comparative assessment.
Overall, the English grammar and spelling are good, but there are some phrases or word choices that should be revised before publication. These are called out below.
Abstract:
Line 12 Grammar “…totally not affected…” This claim is not substantiated by the paper. I recommend using the word “unaffected”.
Line 13 Grammar “…thanks to the proper design…” This too appears to overstate what has been demonstrated in the paper. Simply delete and rewrite this portion of the sentence to read “…and the result of state estimation is unaffected by sensor faults when using the reduced-order sliding mode system.”
Line 15 Grammar “…Unlike those using passive FTC ideas,…” replace the word ‘ideas’ with ‘concepts’.
Introduction:
I recommend that the authors add a graphic to illustrate and present the sliding mode observer concept to the reader.
Line 73 Grammar Again, the phrase “…totally not affected…” which should be replaced with the word “unaffected”.
Line 81 Grammar strike the word ‘good’ from “…nonlinear model of a civil aircraft turbofan engine show good robust fault reconstruction…”
Sensor Fault Diagnosis:
Clear. No comments.
Fault Tolerant Control Based on State Estimation
Line 188 Grammar “…designed observer…” should be “design observer”
Lines 199-200 “It’s clear that the augmented system is not affected by sensor faults.” Actually, this wasn’t clear. A few sentences that summarize the key points or aspects of the equation set that make this so would be helpful. Reference back to the relevant governing equations as needed.
Simulations
Line 234 How was Figure (6a) constructed?
Line 248 Grammar “…scheme is compatible with any existed feedback controller without any controller reconfigurations…” I recommend grammatical change of “existed feedback controller” to “existing feedback controller”
Line 250 Grammar “…thanks to the proper design of reduced-order sliding…”. Again, recommend deleting the phrase ‘thanks to the proper design’ or revising to be more objective.
Lines 253-255 Grammar “…Meanwhile, overshooting in thrust or fuel flow existed in last subsection is avoided here…” Suggest revision so that the phrase reads “…Meanwhile, overshooting in thrust or fuel flow occurred in the last subsection, it is avoided here which represents a major improvement over the prior method.…”
Conclusions
Line 262 Grammar Again, the phrase “…totally not affected…” which should be replaced with the word “unaffected”.
Line 266 Grammar Consider replacing the word ‘ascendancy’ with ‘improvement’ or ‘advancement’.
Author Response
Dear Reviewer:
We deeply appreciate the your comments and suggestions, which are valuable in improving the quality of our manuscript. Detail responses can be found in the enclosure PDF. Should you have any questions about our work, please contact us without hesitate.
Xiaodong Chang
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The paper considers fault estimation and control reconfiguration for aircraft engines.
The idea is to apply a second order sliding mode observer to estimate simultaneously both state and fault.
There are several issues with the paper. Foremost, it is not clear which is the authors' contribution. The theoretical issues seem to be standard (the computation of the sliding mode observer estimation error) and the discussion about the LQR control. If the simulation part is the main contribution, then this should be clearly stated.
Another issue is that I do not see how the particularities of the aircraft engine influence the theoretical part (especially the fact that the engine dynamics are nonlinear whereas the sliding mode uses generic linear dynamics).
I have several other remarks:
In eq. (1) you give a linear, time invariant system. Does this truly correspond to an aircraft engine ? I would have expected a nonlinear dynamic. At line 93 you say that a fault f_i = 1 means total sensor failure but you define it as an additive fault in eq. (1). I would have expected that a total fault means y_i=0 which I do not see happening when replacing f_i=1 in eq. (1). You claim that the fault occurrence does not affect the state estimation. Using a sliding mode observer you arrive at an output error which is zero in a finite time but still, you still have a transitory interval in which the error is not zero (and probably the effect of the fault is not zero). Does this affect the scheme ? At line 217 I see a fault profile which ranges from 0 to -90 but earlier in the paper you said that the variables appearing (including the fault) are normed to stay inside the interval [0,1]. Nowhere in the paper (not even in the simulation part) do you give numerical values for your state matrices. How is the scheme particularized for an aircraft engine, it seems completely generic. Furthermore, at line 206 you mention that the scheme is applied to a nonlinear model. How do you then obtain the linear representation defined in eq. (1) ? Do you linearize the system? If so, in which functioning point ? Figure 6a) would be more useful if you also plot the real fault such that we can compare between the real signal and its estimation. There are some grammar and spelling errors: at line 12, I would not use "totally not affected" but rather "completely unaffected" at line 26, "leads fault handling" does not sound correct at line 77: "in proposed" should be "in the proposed" at line 230: the subsection title appears to be incorrect at line 248: "any existed" should be "any existing"
Author Response
Dear Reviewer:
We deeply appreciate the your comments and suggestions, which are valuable in improving the quality of our manuscript. Detail responses can be found in the enclosure PDF. Should you have any questions about our work, please contact us without hesitate.
Xiaodong Chang
Author Response File: Author Response.pdf
Reviewer 3 Report
The article is describing an new idea control system of an turbofan aircraft engine. In the introductions the authors well define the motivations: lowering the mass of an aircraft and therefore fuel consumption. The study is continuation of previews work. Author also list the relevant references of previous works done on the subject. In the next sections authors describe their work in some more detail: first by developing an mathematical model and afterwords by showing the results of the simulation of the new idea and compering it with the old one. Conclusions seem to support the results.
The introduction is well written and has all the necessary elements that the article should have: it describes the problem, lists the previous works, states the motivations and describes what is the novelty. The sections describe the model in what at least looks as a solid mathematics model (which I unfortunately can not judge in more details, not being an expert in control systems). The results that seem to support what the model states and also include comparison with previous works. Test is written in good English and seems to describe the problem and results in a compact and precise manner. The article seems to be aimed at a narrow reader base that are experts on a control systems.
Author Response
Dear Reviewer:
We deeply appreciate the your comments and suggestions, which are valuable in improving the quality of our manuscript. Should you have any questions about our work, please contact us without hesitate.
Xiaodong Chang
Round 2
Reviewer 2 Report
The authors have made a clear effort to reply to my remarks.
As it stands, the paper appears to be self-contained and easier to read than in the previous version.
Regarding your Q2 reply: It clarifies my previous question about the linear model employed. I would have been happy with only the nonlinear equations which describe the nonlinear model and the static point at which they are linearized.
