Application of Evolutionary Optimization Techniques in Reverse Engineering of Helical Gears: An Applied Study

Round 1

Reviewer 1 Report (Previous Reviewer 2)

The authors in the re-submitted version now addressed most of my comments satisfactory, except:

 1) Parametric statistical t-test has been applied now. But, the authors didn't check (or at least describe) if the following preconditions hold: normality and homoscedasticity. The results from the metaheuristic literature show that a parametric statistical analysis could not be appropriate, especially when we deal with multiple-problem results. For more information, see:

J. Derrac, S. Garcia, D. Molina, F. Herrera. A practical tutorial on the use of nonparametric statistical tests as a methodology for comparing evolutionary and swarm intelligence algorithms. Swarm and Evolutionary Computation 1 (2011) 3-18.

2) The authors still claim in Conclusion: "This research proposes a new methodology for RE of helical gears by meta-heuristic techniques." This is an exaggeration. What is a new methodology? It is a mere application of known methods. Do the authors correctly understand the difference between a methodology and a method? Therefore, the authors should tone down such claims.

 

Typo:

on convergence speed of algorithms can be proposed can be proposed.

->

on convergence speed of algorithms can be proposed.

 

Author Response

Please see the attached file

Author Response File: Author Response.pdf

Reviewer 2 Report (New Reviewer)

The paper is easy to follow but lack of precision in some
sections. Particularly in GWO and WOA explanation. Some doubt arise:

 - There are four groups in GWO: alpha, beta, delta and omega. Why the    last group are not mentioned in the algorithm?

 - How many wolf are in each group?  - In (12) it seems A and D are vectors; what kind of product refers    the dot operator?

 - in (13) the dot operator is used for scalar multiplication and a product between two vectors.

 - What does mean gamma in (17)?

Similar questions arise  for WOA.


There are some typos for example unnecessary closed parentheses in 270 and 285.

Author Response

Please see the attached file.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report (Previous Reviewer 2)

My comments have been addressed, and the paper can be accepted now.

Author Response

Please see attached file.

Author Response File: Author Response.pdf

Reviewer 2 Report (New Reviewer)

Authors modified eq. (12) substituting the dot operator "." by the outer "x" operator. However the outer operator produce a matrix, not a vector (https://en.wikipedia.org/wiki/Outer_product).  It seems authors does not understand the expressions they use in their own paper.

After a closely examination of the original wolf optimizer article (https://www.sciencedirect.com/science/article/pii/S0965997813001853) I came to the conclusion that the dot operator between two vectors is defined by:

Let v = [v1 v2, ..., vn] and w=[w1 w2, ..., wn] two vectors then v.w = [(v1)(w1), (v2)(w2), ..., (vn)(wn)].

I came to this conclusion because authors of the original paper (which is not cited) thoroughly explain the expressions and above all they provide algorithm source code.

Hence I ask the authors cite the original PSO, GWO and WOA papers. But above all provide or refer the exactly source code they use for these algorithms. Not doing so make the paper not understandable.

Author Response

Please see attached file.

Author Response File: Author Response.pdf

Round 3

Reviewer 2 Report (New Reviewer)

Authors mention the use MatLab for the algorithms. However this is not
precise enough for an interested reader reproduce the results. Are
there matlab tools for all algorithms employed? Does they
developed their own software on matlab language? In the first case they
must specify what tools exactly they use; in the second case they must
provide the source code.

Author Response

Please see the attached file.

Author Response File: Author Response.pdf

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

According to my reviewing, I see this paper is not relevant to be published in your journal due to the following issues:

·        The paper is not presented any contributions. For instance, regarding the optimization process, the authors employed three existing algorithms: GWO, PSO, and WOA to optimize the investigated problem; what is new??, is it novel to apply a met heuristic algorithm without any improvements to solve an existing problem??.

·        In addition, the comparison section is so poor because the authors did not compare the adapted algorithms to some of the recently-published state-of-the-arts to affirm that one of these is the best alternative to this problem.

Author Response

Dear Reviewer 1,

I hope this email finds you well.

We want to thank you for reviewing our paper.

Please see the attached file as the Reviewer 1 Response Letter.

Best regards,

Dr Mohammadreza CHALAK QAZANI.

Author Response File: Author Response.pdf

Reviewer 2 Report

The paper applies several metaheuristics (GWO, WOA, PSO, and GA) to solve the reverse engineering of helical gears. Metaheuristics have been used to compute various gear design parameters (normal module, normal pressure angle, addendum modification, and helix angle). As such, this work is a mere application of standard metaheuristics. The authors are even not using some advanced techniques in metaheuristics (e.g., adaptive or self-adaptive parameter control, variable population size, long term memory assistant). Therefore, the novelty and significance of this work are rather low. I am urging the authors to apply some advanced techniques in their revised version.

1) The authors wrote: “reverse engineering (RE) is an approach that consists of digitizing a real component to create a numerical or virtual model 33 [1].” The recent buzzword is “digital twins”. In the related work comparison, usage of digital twins was completely omitted.

2) Related work on design parameter optimization is weak.

3) The authors wrote: “Some of the most popular meta-heuristic optimization methods with successful applications reported in engineering optimization problems, such as GWO, WOA, PSO, and GA, were applied to solve the introduced RE problem.” How was popularity measured? I am not aware of any Systematic Literature Review (SLR) or Systematic Mapping Study (SMS) which supports this claim. Moreover, it seems that the authors are not aware of some critics of GWO.

4) Tables (e.g., Table 1) should not be split over two pages.

5) How t_{max} in Eq. 22 was determined when termination based on a specified error has been used? Namely, under such termination condition, the maximum number of iterations or maximum fitness evaluations are unknown.

6) Not all control parameter settings are mentioned. As such, this work is not repeatable.

7) It seems that metaheuristics control parameters have been arbitrary selected with trial&error approach. Have any tuners been used (e.g., CRS-Tuning, F-Race, REVAC)? At least some discussion is in order.

8) The authors wrote: “Population size for all algorithms was selected equal to perform a fair comparison of accuracy and convergence speed of algorithms.” But, the population size of PSO is 40 (Table 2), and for GA is 32 (Table 3). Hence, the population size for PSO and GA are not equal, whilst population size is not specified for GWO and WOA. Moreover, equal population size does not guarantee the fairness of comparison.

9) In the second scenario, GA didn’t find a solution. But, some results are still reported in Table 9.

10) Algorithm outperformance is not shown by NHST (Null Hypothesis Significance Testing). Is the outperformance of the proposed approach statistically significant?

11) The authors wrote: “This research proposes a new methodology for RE of helical gears by meta-heuristic techniques.” This is an exaggeration. The proposed approach has been used for identifying design parameters in the past for several similar problems.

12) References about metaheuristics are weak. Even inventors of PSO are not cited.

 References used in this review report:

=========================

Dalibor et al. 2022: Generating customized low-code development platforms for digital twins. Journal of Computer Languages, Volume 70, June 2022, 101117

Bano et al. 2022: Process-aware digital twin cockpit synthesis from event logs. Journal of Computer Languages. Volume 70, June 2022, 101121

Vještica et al. 2021: Multi-level production process modeling language. Journal of Computer Languages. Volume 66, October 2021, 101053

Jesenik et al. 2020: Determination of a hysteresis model parameters with the use of different evolutionary methods for an innovative hysteresis model. Mathematics (MDPI), 8.

Jesenik et al. 2018: Searching for soil models' parameters using metaheuristics. Applied Soft Computing, 69, 131-148.

Jesenik et al. 2020: The Use of Evolutionary Methods for the Determination of a DC Motor and Drive Parameters Based on the Current and Angular Speed Response. Mathematics (MDPI), 8, 1269.

Panić et al. 2020: Improved Initialization of the EM Algorithm for Mixture Model Parameter Estimation. Mathematics. 2020; 8(3):373

Camacho Villalon et al. 2020: Grey Wolf, Firefly and Bat Algorithms: Three Widespread Algorithms that Do Not Contain Any Novelty. ANTS 2020, LNCS 12421, pp. 121–133, 2020.

Aranha et al. 2022: Metaphor‑based metaheuristics, a call for action: the elephant in the room. Swarm Intelligence, 16:1–6 

LaTorre et al. 2021: A prescription of methodological guidelines for comparing bio-inspired optimization algorithms. Swarm and Evolutionary Computation 67 (2021) 100973

Ravber et al. 2022: Maximum number of generations as a stopping criterion considered harmful. Applied Soft Computing, Volume 128, October 2022, 109478.

Derrac et al. 2011: A practical tutorial on the use of nonparametric statistical tests as a methodology for comparing evolutionary and swarm intelligence algorithms, Swarm Evolut. Comput. 1 (1):3–18, 2011.

Author Response

Dear Reviewer 2,

I hope this email finds you well.

We want to thank you for reviewing our paper.

Please see the attached file as the Reviewer 1 Response Letter.

Best regards,

Dr Mohammadreza CHALAK QAZANI.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

The authors only slightly revised the paper, and many of my previous comments (see the review report from the first round) have not been addressed. The most important comments which must be addressed are:

 1) The authors wrote: “Confirmatory calculations and experiments reveal the remarkable efficiency of Grey Wolf Optimization and Particle Swarm Optimization techniques in reverse engineering of helical gears compared to other techniques and obtaining influential gear design parameters.” But, GWO and PSO outperformance is not shown by NHST (Null Hypothesis Significance Testing). Is the outperformance of GWO and PSO statistically significant? If GWO and PSO performance is remarkable, then they must be statistically significantly better. 

2) There is no discussion on tuners (e.g., CRS-Tuning, F-Race, REVAC) and how this work can benefit using them.

3) The authors are not using some advanced techniques in metaheuristics (e.g., adaptive or self-adaptive parameter control, variable population size, long term memory assistant). Therefore, the novelty and significance of this work are rather low. These approaches are not even discussed as future work.

4) The authors wrote: “This research proposes a new methodology for RE of helical gears by meta-heuristic techniques.” This is an exaggeration. The proposed approach has been used for identifying design parameters in the past for several similar problems. Therefore, the authors should tone down such claims.

5) References on Digital Twins are weak.

Typos:

Introduced by Eberhart and kennedy

->
Introduced by Eberhart and Kennedy

Author Response

Please see the attached file!

Author Response File: Author Response.pdf

Round 3

Reviewer 2 Report

I am disappointed with the authors' response. They are not willing to improve the paper despite the fact that improvements will not take much effort. Note that one of the main goals of reviewers is to suggest improvements (besides checking the soundness and significance of the proposed approach). I am briefly stating again issues which must be improved (for details, see my previous reports):

1) The outperformance of the proposed approach has not been shown by NHST (Null Hypothesis Significance Testing). The authors now claim that there is no need to perform NHST since some papers exist where NHST was not applied. But, mistakes from previous research should not be repeated. They are many more research works where NHST has been applied. Seems that the authors are unable to perform simple statistical analysis.

2) There is no discussion on tuners (e.g., CRS-Tuning, F-Race, REVAC) and how this work can benefit using them. The authors responded that tuners were not used, and hence there is no need to discuss their importance.

3) The authors wrote: "This research proposes a new methodology for RE of helical gears by meta-heuristic techniques." This is an exaggeration. The proposed approach has been used for identifying design parameters in the past for several similar problems. Therefore, the authors should tone down such claims.

4) References on Digital Twins are weak.

If the authors are not willing to improve the paper in this round, my recommendation is to reject this paper. As I have already mentioned in my first review, the novelty and significance of this work are rather low. It is a mere application of known methods.