Flexible Flow Shop Scheduling Method with Public Buffer
Round 1
Reviewer 1 Report
very interesting paper. following issues need to be addressed.
the paper needs to be reviewed for the grammatical and spelling mistakes. Section 4.3 -> justification is missing how the decision are made regarding the experimentation; the number of runs (e.g. 30; why 30 runs?); why 500 generations? Problem description and constraints section require improvement. Explanation required for the constraints, ideally an example to go with the explanation to link this with the Gantt chart presented in Figure 4 Figure 4 shows Gantt chart analysis of scheduling results, however, flexible flow shop representation is missing; what is the job sequence? Add those details so that it can be verified if the hard constraints are maintained. The conclusion must be added to the paper; what are the future directions? Limitation of the current approach? Relevance to real-world manufacturing problems. Nothing said regarding the algorithm performance; what is the problem size? impact of the number of machines, routing, number of jobs on the algorithm performance.
Author Response
Point 1: The paper needs to be reviewed for the grammatical and spelling mistakes. 

Response 1: The grammatical and spelling mistakes in this paper have been corrected.
Point 2: Section 4.3 -> justification is missing how the decision are made regarding the experimentation. 

Response 2: I am really sorry. Section 4.3 title of original paper is wrong and it is the same to Section 4.2. Now, the title has been corrected and the new title is Optimization performance testing on the SAA-HNN algorithm.
Content in Section 4.3 is Optimization performance testing on algorithms. Meanwhile, in order to clearly express writing purpose of this section, some parts in Section 4.3 have been changed.
Point 3: The number of runs (e.g. 30; why 30 runs?); why 500 generations? 

Response 3: The purpose to choose 30 simulation experiment times is to calculate the average values of optimizing result during multi-time simulating experiments and to further test the optimizing effect of algorithms. Because Hopfield neural network algorithm and its comparative algorithms cannot ensure that they must get the optimal global solution during optimizations, most optimizing results are the optimal local solutions, and further, optimizing results are more than one. Simulating experiments are needed for many times to get the average values in order to determine average optimizing performance of algorithms. Meanwhile, 30 times simulation experiments are because the scale of experiment operation times is enough and most referencing literatures in related fields choose 20-30 operating times as simulation experiment times. Therefore, this paper chooses 30 as operating times of simulation experiments.
After the optimizing performance of Hopfield neural network algorithm is tested, we discover because energy function of this algorithm is monotonically decreasing, the optimization range of the algorithm is narrow with a fixed optimization direction. This leads evolving process of Hopfield neural network algorithm stagnates between 100 iteration and 200 iteration. In order to observe difference of optimizing results between comparative algorithm and Hopfield neural network algorithm, about twice evolutional generations are chosen, that is, 500 iteration is taken as the maximum iteration time of algorithms.
Point 4: Problem description and constraints section require improvement.

Response 4: Problem description and constraints section have been correspondingly improved. The new public buffer constraint has been set, that is, constraint (17) has explained this constraint meaning.
Point 5: Explanation required for the constraints, ideally an example to go with the explanation to link this with the Gantt chart presented in Figure 4 Figure 4 shows Gantt chart analysis of scheduling results, however, flexible flow shop representation is missing; what is the job sequence? Add those details so that it can be verified if the hard constraints are maintained.

Response 5: In order to clearly understand and describe flexible flow ship representation, job sequence and constraints function, Gantt chart of Figure 4 has been corrected and constraint effect has been explained in the analysis of this figure. Because of length limitation in this paper, only some important constraints have been described and other constraints will be explained only in this response. As is shown:
Based on next chart(Specific examples can observe processing path of workpiece in blocks through line connection), it is clear to see in Figure 1:
Restricted by constraint(7), the completion time of workpiece in the stage is equal to the sum of its start time and standard processing time in the stage .
Restricted by constraint(8), workpiece needs to complete the processing task of the current stage before the processing task of the next stage.
Restricted by constraint(9), time for the workpiece to enter the limited buffer cannot be less than the completion time of this workpiece processed at the previous stage .
Restricted by constraint(12), the workpiece to leave limited buffer cannot be less than the time for the workpiece to enter limited buffer .
Restricted by constraint(13), the workpiece in public buffer that should have entered limited buffer to leave the public buffer that should not be less than the time for it to enter public buffer .
Restricted by constraint(15), at any time, the sum of workpieces contained in the to-be-processed collection is ≤ the maximum buffer capacity of the public buffer.(This constraint can be seen in Figure 2)
Figure 1 and Figure 2 are in the attachment. Please see the attachment.
Point 6: The conclusion must be added to the paper; what are the future directions? Limitation of the current approach? Relevance to real-world manufacturing problems. 

Response 6: I am really sorry that the title of Section 6 is wrong and the content in Section 6 is this paper’s conclusions. Therefore, the title of Section 6 is corrected as Conclusions.
Based on problems of real manufacture process, current approach limitation and future directions are added in Section 6.
Point 7: Nothing said regarding the algorithm performance; what is the problem size?impact of the number of machines, routing, number of jobs on the algorithm performance.
Response 7: I am really sorry. Section 4.3 title of original paper is wrong and it is the same to Section 4.2. Now, the title has been corrected and the new title is Optimization performance testing on the SAA-HNN algorithm.
Two simulation experiments of new larger-scaled data are operated and the purpose is to expand simulation data scale to a new magnitude so as to more effectively test optimization effect of algorithms.
Add related explanations of the example expression form in Section 4.3 and this more clearly describes data scale of this examples.
The impacts of the number of machines, routing, number of jobs on algorithm performance are further analyzed and described.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The authors solve a very difficult problem. The flexible flow shop scheduling is solved with Public Buffer(FFSP-PB): a new method combining the simulated annealing algorithm-based Hopfield neural network algorithm(SAA-HNN) and local scheduling rules. Successful experiments are made
A link with the running model, at least the executable part, and some images of the running model could be a plus.
LaTex to edit scientific papers
Editing issues: - comma after words, spaces between words
Author Response
Point 1: A link with the running model, at least the executable part, and some images of the running model could be a plus. 

Response 1: The link of running model is shown as:
Link: https://pan.baidu.com/s/1FD2CejXJ2ghreGoDFkqRUg
Extracted code： pj8h
Point 2: LaTex to edit scientific papers
Response 2: I have not learnt LaTex and response time limitation of review is short so LaTex cannot be temporarily used to re-typeset but some mistaken formats in this paper are corrected.
Point 3: Editing issues: - comma after words, spaces between words 

Response 3: The editing issues of your point in this paper have been corrected.
Author Response File: Author Response.pdf
Reviewer 3 Report
The manuscript titled “Flexible Flow Shop Scheduling Method with Public Buffer” by Zhonghua Han, is clearly written, This study explain the problem with FFSP-PB and to address the problem the author proposed an algorithm for global optimization based on simulated annealing, Since standard Hopfield Neural Network Algorithm easily falls into local extremum and is difficult for continuous evolution. Due to high probability of falling into local extremum the proposed algorithm has the potential to accept poor fitness solution and further expand its search scope during asymptotic convergence. This study also designs multiple local scheduling rules to control the moving process which has the potential to improve the efficiency of the workpiece transfer. The manuscript can be accepted as it is.
Author Response
Thank you for your approval of this article.
Reviewer 4 Report
I want to congratulate the authors for the excellent work. In my opinion, the paper is in acceptable form as it is.
This problem addresses the flow shop scheduling problem and proposes a new method, based on a known optimization tehnique, the simulated annealing, the Hopfield neural network algorithm and local scheduling rules. The Introduction part is very well-written and gives the reader a good idea about the importance of the topic studied, and stresses how this work differs from other papers on the topic.
The problem is very-well defined in Section 2 and the model parameters are rigorously defined.
The explanations regarding the improvements offered by the typical HNN algorithm based on the fact that the energy function is monotonically decreasing is well-binded with the idea of using the simulated annealing (SAA) with HNN.
The proposed scheme was compared to numerous other schemes and its improvements are clearly described and support the analysis presented.
In my opinion, this paper is very-well written (i have seen no serious language mistakes) and the References section includes new and well-established schemes, something that shows to me that the topic is of great interest.
Again, congratulations to the authors,
Author Response
Thank you for your approval of this article.
Round 2
Reviewer 1 Report
I would like to thank the authors to address all the comments with adequate information.
