Optimization of Sour Water Stripping Unit Using Artificial Neural Network–Particle Swarm Optimization Algorithm

Round 1

Reviewer 1 Report

I recommend the manuscript for publications after these major revisions

1. The manuscript's English language should be improved by the authors.
2. Introduction needs to enrich the readers with recent papers related to your work, which gives the reader a clear visual of the gap that those studies have not address and covered by this study.
3. The results are not enough to defend the importance of work, please add more quantitative outcomes.
4. The reviewer couldn't see the significant objectives of the current work. It would be better if you add a separate section with Key objectives heading.
5. Why this study is more important and how it could contribute to solving the relevant problem.
6. The methodology part doesn't clear; it may not catch the attraction. Try to make it more concise and briefer.

Author Response

Dear reviewer,

Thanks a lot for handling this manuscript.  In the attachment, the detailed point-by-point replies based on the comments and suggestions are attached.

Author Response File: Author Response.docx

Reviewer 2 Report

There is a need to improve the paper, at present it is not well written.

1)      Try to avoid the abbreviation in the abstract, if not necessary

2)      In the introduction section, authors should deliver to discuss how references had been selected and the limitations of the conducted research study.

3)      Although the field of study is on meta-heuristic algorithm, but this issue has not been considered much in the writing of the article and has been marginalized. Therefore, it is suggested that in the introduction and in the proposed approach" section, using appropriate references, what is the subject and the necessity of the problem be explained.

4)      Update the introduction section with some articles such as A hybrid PSO-GA algorithm for constrained optimization problems; A hybrid GSA-GA algorithm for constrained optimization problems; Simulation on Supplier Side Bidding Strategy at Day-ahead Electricity Market Using Ant Lion Optimizer. Journal of Computational and Cognitive Engineering;

5)      Mention the objective of the study in the introduction section in bullet points.

6)      Before section 3, authors should put some basic definitions so that readers can well know about the work.

7)      Improve all the figures with better resolution.

8)      There are many errors seen in the work related to the presentation as well as concept.

9)      Add the advantages of the proposed work with a separate section.

10)  The conclusion is not precise. The key findings and its implementation potential (in practice) is missing. Clearly, identify its academic contributions also. Limitations are not mentioned.

11)  Point out the limitation of the study in the conclusion section.

 

12)  Future research work should be added with the article such as Research on Robot Path Perception and Optimization Technology based on Whale Optimization Algorithm. 

Author Response

Dear reviewer,

Thanks a lot for handling this manuscript.  In the attachment, the detailed point-by-point replies based on the comments and suggestions are attached.

Author Response File: Author Response.docx

Reviewer 3 Report

The authors optimize a chemical process of the acid water vapor stripping unit (modelled in the chemical engineering software Aspen Plus).
The authors combine the ANN approach and a PSO approach in order to achieve this. The authors use a variance analysis method (Sobol) to investigate the features of the system.
There is no doubt that the topic at hand is relevant, however, the manuscript needs major improvements in order to be published.

The English in the manuscript needs improvement with further proof reading. Some sentences need to be rewritten in order to make more sense - there are not line numbers in the manuscript hence it is hard to point exactly where.

There are some unexplained abbreviations in the abstract (BP, RBG, GRNN).
The introduction section feels a bit weak with a past studies review. The authors state that "many scholars have carried out a large number of relevant studies and achieved certain results" which is a vague statement.

In the Process simulation section the authors state that "Aspen Plus is the best choice to simulate and establish the mathematical model in this study", however, this is just a statement without proper numerical validation. Furhtermore, please fix "stimulation" to "simulation".

Figure 1 - The diagram could be larger and the diagram markers should be thoroughly explained.
The paragraph below Figure 1 is too vague, the authors should specify more on this -- which operating parameters were changed? What are the groups od collected  data? Which were the inputs/outputs? This all needs to be specified for reproducibility purposes.

Section 2.2. - ANN having short computing time is highly debatable, maybe inference, however, training can last a long time. The authors should probably denote RBF as RBFNN as RBF means radial basis function, however, they used RBFNN for their study. Same comment for the BP neural network.

Section 3 - Why was the train/test/validation split of data selected like that and not through a random shuffle? Furhtermore, the authors should do a cross-validation analysis in order to make sure that their model is robust in its prediction. Each prediction given by the authors should have a standard deviation obtained through cross validation (KFold). Furthermore, the authors should present loss curves of the best performing ANN model in order to show that their model avoids overfitting.

Figure 4 - Please explain in the caption what is being shown at each of the subfigures.

Section 3.3. - Please specify what were the PSO algorithm parameters e.g. cognitive, social factor, inertia etc. Furthermore, the authors should repeat the PSO optimization procedure several times in order to show that they haven't converged in local minima. The average value of fitness (Energy consumption) through several runs should be reported.

Author Response

Dear reviewer,

Thanks a lot for handling this manuscript.  In the attachment, the detailed point-by-point replies based on the comments and suggestions are attached.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

The authors have satisfactorily addressed most of my concerns

Author Response

Dear editor and reviewer,

Thanks a lot for handling this manuscript again. In the attachment, the detailed point-by-point replies based on the comments and suggestions are attached.

Author Response File: Author Response.docx

Reviewer 2 Report

The paper still needs to be improved. A lot of errors and unclear things given.

1)       The first line of the abstract “To further reduce….”, What reduce..?? It is not good sentence to start with this.

2)       It is unclear why there is a need to optimize and use PSO rather than other algorithms such as GA, TLBO, HHO etc

3)       Heading of section 2: Method is not good. In this section it is not proposed the method.

4)       Update the literature with some article on PSO and GA to show the importance of PSO algorithm.

5)       Comparative analysis is missing in the study.

6)       Add the advantages of the method in a separate paragraph to clearly distinguish from the existing ones.

 

7)       Conclusion should be rewritten. At present, it is same as that of Abstract… 

Author Response

Dear editor and reviewer,

Thanks a lot for handling this manuscript again. In the attachment, the detailed point-by-point replies based on the comments and suggestions are attached.

Author Response File: Author Response.docx

Reviewer 3 Report

The authors have made major improvements to the manuscript.

Author Response

Dear editor and reviewer,

Thanks a lot for handling this manuscript again. In the attachment, the detailed point-by-point replies based on the comments and suggestions are attached.

Author Response File: Author Response.docx

Round 3

Reviewer 2 Report

accepted