An Optimized Fuzzy Logic Control Model Based on a Strategy for the Learning of Membership Functions in an Indoor Environment
Round 1
Reviewer 1 Report
A really nice manuscript to read in terms of theory and application. At times the text was a little verbose, giving much more to read than is probably necessary, e.g. pages 1, 2 and 3 are a lot to read...……...a little editing on the length would make the reader enjoy it a bit more.
Author Response
Manuscript ID: electronics-384029
Title: An Optimized Fuzzy Logic Control Model Based on Learning of Membership Functions Strategy in Indoor Environment
Authors: Muhammad Fayaz, Israr Ullah, and DoHyeun Kim
Dear reviewer,
Thank you for your instructions to revise the manuscript. Please Attached herewith find our revised manuscript of the above referred paper for your consideration and publication in Processes Journal. We have thoroughly revised and modified the manuscript according to the comments.
A really nice manuscript to read in terms of theory and application. At times the text was a little verbose, giving much more to read than is probably necessary, e.g. pages 1, 2 and 3 are a lot to read...……...a little editing on the length would make the reader enjoy it a bit more.
v We really appreciate your valuable time and the encouraging and supportive feedback.
v Introduction and related sections are revised and some redundant and un-necessary text is removed to improve its readability according to your suggestion.
v We have also removed some unnecessary figures and reduce overall size of the paper to make it more readable.
Please contact me if you have any questions on the revised manuscript.
Best regards,
Muhammad Fayaz
Author Response File: 
Author Response.pdf
Reviewer 2 Report
General remarks
I would like to appreciate very much this interesting and practice oriented paper.
In fact, the Authors proposed a cooperative neuro-fuzzy system in which neuronal network is used for preparing and updating fuzzy membership functions used in fuzzy indoor car temperature controller. The architecture of this class of systems is known since many years.
I am not convinced by obtained results. Firstly, the enhancement of the root means square error in comparison to conventional fuzzy logic controller (line 354) is so small, that taking into account some portion of uncertainty is seems illusory.
Moreover, the results reported in this paper refer to the simulations only. However, the simulation results reported in this paper are not verified experimentally. This does not allow evaluating the obtained results completely. In addition, the comparison obtained the control quality indices with these obtained by means of classic PID controller are not given too. This makes paper incomplete.
Heat and cooling systems are in fact a higher order lag systems. Therefore, the effective temperature control system has to some extent exhibit predictive features in order to cope with required control quality indices. However, this requires not only the knowledge of the current ambient and required indoor temperature but also some knowledge about previous behaviour (dynamics) of the controlled system. I am afraid, that control system which does not take these factors into consideration does not guarantee optimal control quality indices. In fact, the applied fuzzy rule base given in Tab. 2 relates to a fuzzy nonlinear P controller. Is really, the application of a P controller the best solution for a temperature control?
Novelty
Novelty of the paper should be expressed explicitly. It is not clear if the novelty of the paper rely of the application of ANN in the process of determining of fuzzy membership function shapes or application of fuzzy logic controller for indoor car temperature control. The first issue seems not to be particularly new.
Motivation of the paper is understandable. The paper is application oriented. However, both issues (novelty and motivation) are not sufficiently clear expressed and/or underlined in the text.
Contribution
Contribution of the paper should be defined.
Language
I am not a native speaker, but I feel that the style and grammar of the paper is not perfect.
References
Line 217 – incorrect reference. This reference points out a simulation environment. It should refer to an ABS model.
Details
1) Line 9.
Term of “Mamdani fuzzy logic” sounds a little bit strange. Probably the Authors mean fuzzy inference based on Mamdani’s implication which is in common for a wide class of fuzzy systems.
2) Line 11.
Cit: “The most important issue is accurate determination of suitable membership functions (MFs) distribution.” I suggest to extend this sentence by adding two additional factors: shape and boundaries of MF in the universe of discourse.
3) Line 43
Typo.
4) Line 45
Cit.” It is used to uncertaint. “ Please correct the sentence.
5) Line 79
What does it mean? “Accurate number of membership functions ... ”
6) Line 113
Typo.
7) Line 147
Cit: “In conventional fuzzy logics methods membership functions is carried out randomly”. Generally, it is not true.
8) Line 158
Cit:” In traditional fuzzy logic controller, there no proper mechanisms of membership functions determination”. But there are available some rules and templates how to handle this problem.
9) Line 205
The difference between environmental temperature and required temperature in Fig. 3 is not D.
10) Line 205/206
Cit:”The output of the is the membership function set for FLC.” Please complete the sentence.
11) Line 215
Word “below” is not necessary here.
12) Line 222
Typo
13) Line 232
Please check formula (8) and (9).
14) Line 237
Please correct the sentence.
15) Line 247
Please correct the sentence.
16) Line 248
Please correct the sentence.
17) Formula (13)
Conflict of symbol “a” with formulae (3) and (4).
18) Formula (14)
Formula should be rewritten in order to be readable.
19) Fig. 5
Poor quality of the Fig. 5.
20) Line 281
What does it mean “accuracy of fuzzy logic controller”?
21) Line 310
Cit.: “Standard deviation of the input variable difference which is equal to 0.7.”. Because it is standard deviation of temperature – it should have physical unit.
22) Line 336
Typo.
23) Line 337
Font.
24) Line 349
Typo.
25) Line 350
Please insert comma.
26) Line 352
“suggested” ?
27) Line 381
Typo
28) Fig. 10.
Temperature is in [°C] not in (c).
How relates “Time (15m)” to description of x-axis?
Author Response
Manuscript ID: electronics-384029
Title: An Optimized Fuzzy Logic Control Model Based on Learning of Membership Functions Strategy in Indoor Environment
Authors: Muhammad Fayaz, Israr Ullah, and DoHyeun Kim
Dear reviewer,
Thank you for your instructions to revise the manuscript. Please Attached herewith find our revised manuscript of the above referred paper for your consideration and publication in Processes Journal. We have thoroughly revised and modified the manuscript according to the comments. Detailed responses to the comments are listed below point by point.
We appreciate the valuable comments. Line number are provided in responses to the comments (if needed) and highlighted text where changes are made in the paper text.
Author Response File: Author Response.pdf
Reviewer 3 Report
Line 12, 61, 79, 208 and more: In scientific studies no personal phrases are used, e.g. I, we etc...
In Figures 3 and 4 you did not explain all the variables (signals).
Line 205: Where is the variable D in figures 3 and 4 - required temperature?
Formulas (1), (2), (20) - why do you use a vector product?
Formula (10) - what does linear (x) mean? This is a typical linear function?
Figure 5 looks strange, it is not technical / scientific. In addition, the descriptions are too large. Maybe you should use abbreviation to describe the membership function, e.g. Extremely very low - EVL as in table 2.
Line 337 and 338: too large font.
Some of the formulas are in italics and some are normally.
Line 344: should be Equation (21).
Figure 7: Figures a) and b) are the same.
Figures 8 a) and b), Figures 9 a) and b): It seems to me that these drawings are unnecessary, most Matlab users know this.
Figures 10 and 11: I do not understand these diagram completely. First of all, the tempreature should be in degrees Celsius, unit oC. Horizontal axis: Is the time given in 15 minute intervals? If this is the case, the whole process takes 15min * 90 = 1350min? Why is the required temperature decreasing and the current temperature increasing? Here I can not see any control process. This is what you showed in Figures 12. What is the Fan Speed unit? - (rpm)?
It seems to me that you should review the manuscript one more time, especially the penultimate chapter.
Author Response
Manuscript ID: electronics-384029
Title: An Optimized Fuzzy Logic Control Model Based on Learning of Membership Functions Strategy in Indoor Environment
Authors: Muhammad Fayaz, Israr Ullah, and DoHyeun Kim
Dear reviewer,
Thank you for your instructions to revise the manuscript. Please Attached herewith find our revised manuscript of the above referred paper for your consideration and publication in Processes Journal. We have thoroughly revised and modified the manuscript according to the comments. Detailed responses to the comments are listed below point by point.
We appreciate the valuable comments. Line number are provided in responses to the comments (if needed) and highlighted text where changes are made in the paper text.
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
General remarks
The paper is significantly improved in comparison to previous version.
The general idea of a temperature control system proposed schematically in Fig. 1 is quite clear. However, this control scheme is much more complicated in comparison to the classic temperature control schemes. Therefore, I suppose that the Authors should justify this complication and clearly show in example or at least specify the expected advantages of the proposed control scheme over the classic approaches.
The laboratory model of the temperature control systems is easily available in public domain. I do not understand why the Authors did not try to implement developed control scheme in order to verify the approach, recognize its advantages and disadvantages in respect to classic approaches. Please explain this issue.
Moreover, the introduction is in part devoted to the problem of energy savings. This issue is not discussed later in this paper in respect to application of developed control algorithm. What energy savings are expected?
Figs. 1, 2 deliver almost the same information as Fig. 1. The only difference is the learning data source. What data are used for learning? I can suppose that archive temperatures from outdoor temperatures. Additionally, the feedback from real system is used. In this case the system is learned to control required temperature based on historic data.
Editorial remarks
The paper should be edited more carefully. The second revision is much better readable compared to previous version. However, some parts of the paper or the same messages are repeated in the paper. See details in Detailed Remarks section. Please make appropriate revisions.
Conclusions
Please state drawbacks of the proposed approach.
Detailed remarks
a) Line 45. Style of the sentence.
b) Line 60. Style.
c) Line 65. Frankly speaking Mamdani fuzzy logic is not strictly controller.
d) Lines 142,143, 144 and 153,154,155 repetition of the same message.
e) Fig. 1. What role plays the connection between User Requirement and Learning Algoritms blocks. The arrow is omitted in this connection.
f) Lines 162,163 and 166,167. Repeated information. Please make fusion.
g) Line 171. Cit.” The output of the fuzzy logic is the required power for the fan.”
I propose to use to change this sentence eg. “The defuzzyfied output of inference machine of fuzzy controller ….”.
h) Lines 190, 191. Repeated information.
i) Line 196. Proptosis?
j) Lines 199..201. Really all inputs to ANN are stated? See Fig. 3.
k) Lines 287, 288. Is it really fundamental purpose of the paper? It is clear that is important to adaptively shape the MFs, particularly for non-stationary systems. There are a lot of methods developed.
l) Line 317. Such accuracy of standard deviation is not justified.
m) Lines 384..387. Repeated information.
n) Line 416. I understand that here real temperature is not strictly real but simulated.
o) Fig. 8b) please correct visibility of y-axis description,
p) Fig. 9a) Error in caption.
q) Line 436. Please revise sentence.
r) Line 437. Add fan rotational speed.
s) Fig. 9b) Power unit is false.
t) Lines 463..465. The conclusion may strongly depend on tuning results of both coppared controllers.
Author Response
Manuscript ID: electronics-384029
Title: An Optimized Fuzzy Logic Control Model Based on Learning of Membership Functions Strategy in Indoor Environment
Authors: Muhammad Fayaz, Israr Ullah, and DoHyeun Kim
Dear reviewer,
Thank you for your instructions to revise the manuscript. Please Attached herewith find our revised manuscript of the above referred paper for your consideration and publication in Processes Journal. We have thoroughly revised and modified the manuscript according to the comments. Detailed responses to the comments are listed below point by point.
We appreciate the valuable comments. Line number are provided in responses to the comments (if needed) and highlighted text where changes are made in the paper text.
Response to Reviewer:
1. The general idea of a temperature control system proposed schematically in Fig. 1 is quite clear. However, this control scheme is much more complicated in comparison to the classic temperature control schemes. Therefore, I suppose that the Authors should justify this complication and clearly show an example or at least specify the expected advantages of the proposed control scheme over the classic approaches.
v In this work, we consider to evaluate the performance of conventional FIS using ANN through learning mechanism. We also compared the results of the proposed scheme with conventional FIS and satisfactory improvement is observed. In the proposed, work we have only considered the conventional Mamdani fuzzy inference method and our ultimate goal was to improve its performance because further, we will use this fuzzy inference method in our work to minimize energy consumption
“The most important advantage of the proposed approach is that it enhances the performance (in term of accuracy) of the conventional Mamdani fuzzy logic method by defining accurate membership functions boundaries using a neural network. The proposed system design is kept flexible”.
2. The laboratory model of the temperature control systems is easily available in the public domain. I do not understand why the Authors did not try to implement developed control scheme in order to verify the approach, recognize its advantages and disadvantages in respect to classic approaches. Please explain this issue.
v As I mentioned earlier that the In the proposed, work we have only considered the conventional Mamdani fuzzy inference method and our ultimate goal was to improve its performance because further we will use this fuzzy inference method in our work to minimize energy consumption. We have previously worked on Mamdani fuzzy inference method to give required power for actuators and we want to improve the performance of conventional Mamdani fuzzy inference method.
v Thanks for your kind suggestion, in future we would consider other control methods in our work.
3. Moreover, the introduction is in part devoted to the problem of energy savings. This issue is not discussed later in this paper in respect to the application of the developed control algorithm. What energy savings are expected?
v We revised the introduction section and removed some text related to energy optimization. We added some relevant text to neural network and fuzzy logic system.
4. Figs. 1, 2 deliver almost the same information as Fig. 1. The only difference is the learning data source.
v Yes, but Figure 1 is the conceptual model and Figure 2 is the detail, and our purpose was to better describe the proposed approach step by step, therefore, we have considered both.
5. What data are used for learning? I can suppose that archive temperatures from outdoor temperatures. Additionally, the feedback from the real system is used. In this case, the system is learned to control the required temperature based on historical data.
v The ANN algorithm is trained on some historical data where the inputs are previous temperature (Tp) from the environment; the previous user required temperature (Trp) and the difference of both (Dp), and their corresponding outputs are the MFs sets.
v The feedback from the real system is the current temperature values, the current required temperature values and difference of both.
à Editorial remarks
Conclusions
§ Please state drawbacks of the proposed approach.
v The proposed learning to control has also some disadvantages, such as it required historical data for training, and also it computation time is high than conventional Mamdani fuzzy logic method due to the learning module.
à Detailed remarks
a) Line 45. Style of the sentence.
v The modification has been done as suggested.
b) Line 60. Style.
v The modification has been done as suggested.
c) Line 65. Frankly speaking, Mamdani fuzzy logic is not strictly controller.
v Yes, you are right but here we term the Mamdani fuzzy logic is a controller because the output of the Mamdani fuzzy logic method is the required power for actuators and according to that power the status of the actuators are controlled. The second reason is that in our previous studies we named the Mamdani fuzzy logic method as a controller and in future, we have intended to use the same method in our previous work, therefore, we named as the controller.
d) Lines 142,143, 144 and 153,154,155 repetition of the same message.
v The repetition has been removed.
e) Fig. 1. What role plays the connection between User Requirement and Learning Algorithms blocks. The arrow is omitted in this connection.
v The missing arrow has been corrected.
f) Lines 162,163 and 166,167. Repeated information. Please make fusion.
v The repetition has been removed as suggested.
g) Line 171. Cit.” The output of the fuzzy logic is the required power for the fan.”
I propose to use to change this sentence eg. “The defuzzyfied output of inference machine of fuzzy controller ….”.
v The modification has been done according to your suggestion.
h) Lines 190, 191. Repeated information.
v The repetition has been removed.
i) Line 196. Proptosis?
v Corrected: Purposes.
j) Lines 199.201. Really all inputs to ANN are stated? See Fig. 3.
v We redraw Figure 3 and stated all inputs in the text.
k) Lines 287, 288. Is it the really fundamental purpose of the paper? It is clear that is important to adaptively shape the MFs, particularly for non-stationary systems. There are a lot of methods developed.
v In the proposed work the tuning of the MFs has been carried in order to get better results as compared to the conventional Mamdani FLC method where the MFs are fixed.
l) Line 317. Such accuracy of standard deviation is not justified.
v Accuracy of standard deviation is not defined because we just observe the standard deviation value that shows the variation in data and according to that variation the numbers of membership functions are defined. As here we have only one input parameter therefore we have only one standard deviation value, in case of many input parameters then different threshold can be set and according to those thresholds the numbers of membership functions can be defined.
m) Lines 384..387. Repeated information.
v According to your suggestions the repeated information has been removed.
n) Line 416. I understand that here the real temperature is not strictly real but simulated.
v Yes, you are right, real temperature is not strictly real but simulated
o) Fig. 8b) please correct visibility of y-axis description.
v The modification has been done as suggested.
p) Fig. 9a) Error in caption.
v The correction has been done accordingly.
q) Line 436. Please revise the sentence.
v The modification has been done as suggested.
r) Line 437. Add fan rotational speed.
v The modification has been done as suggested.
s) Fig. 9 b) Power unit is false.
v The correction has been done according to your suggestion.
t) Lines 463.. 465. The conclusion may strongly depend on tuning results of both compared controllers.
v Modified and added some more information
Please contact me if you have any questions on the revised manuscript.
Best regards,
Muhammad Fayaz
Author Response File: Author Response.pdf
Reviewer 3 Report
All my comments have been taken into account. The article can be published.
Author Response
Dear reviewer,
Thanks for your nice comments.
