Using a Novel Algorithm Based on the Random Vector Functional Link Network and Multi-Verse Optimizer to Forecast Effluent Quality
and Yalei Zhang 2
Round 1
Reviewer 1 Report
The article presents an interesting approach to the subject of the prediction of the effluent quality using ANN. The work is interesting from the standpoint of computational environmental science and has been thoroughly prepared. Results were also statistically evaluated. Prior to publication, I suggest the following minor revisions to the authors:
1.1. Please provide the link (or more information) for the initial influent datasets used so that other researchers could check and reproduce your results.
2 2. Revise the Abstract in order to make it more concise including the information of point 1 above.
3. How did you select the number of hidden layer nodes? Please include this Table in Supplementary data, if possible.
4. Correct minor language errors.
Author Response
We would like to thank you for your careful reading, helpful comments, and constructive suggestions, which has significantly improved the presentation of our manuscript. The changes to the article are described below.
- Please provide the link (or more information) for the initial influent datasets used so that other researchers could check and reproduce your results.
Re: Thanks for your reminder, in line 297 of the main text, we added the download URL for the influent dataset, which also includes a more detailed description of BSM1.
- Revise the Abstract in order to make it more concise including the information of point 1 above.
Re: Thanks for your suggestion, we're sorry about that. To make the abstract more explicit, we tried the following revisions.
Firstly, we changed the sentence “Some critical parameters, biochemical oxygen demand (BOD5) and chemical oxygen demand (COD), are two examples that are challenging to measure in real-time.” to “Reliable effluent prediction is critical in the scientific management of water treatment plants”, since we feel that the prior sentence does not adequately explain the research topic (in lines 16-17).
Secondly, we revise the description of our research process and make an effort to briefly express it, which can be seen in lines 25-29. The revised description is that“The proposed approach is utilized to anticipate real-time effluent data obtained from Benchmark Simulation Model 1(BSM1). The results of the experiments demonstrate that the MVO methodology can successfully find the optimum input-hidden weights and hidden biases of the RVFL model while outperforming the original RVFL and other typical machine learning approaches in all types of influent datasets.”
- How did you select the number of hidden layer nodes? Please include this Table in Supplementary data, if possible.
Re: Thanks for your suggestion. To be honest, the amount of hidden nodes is manually adjusted. We choose the number of nodes from the set of {2n-1,2n,2n+1} and use the one with the minimum error in the end(n is the number of input variables). In reality, there are some published publications that use swarm intelligence approach to determine the ideal number of RVFL network hidden layer nodes. However, we prefer to believe that when the input weights and biases is created at random, the ideal number of hidden layer nodes discovered by this technique may vary. Thank you for reminding us; we will add the following Table in the Supplementary.
|
Influent dataset |
Number of Hidden layer nodes |
RMSE of BOD5 |
RMSE of COD |
|
Dry |
19 |
0.066 |
0.453 |
|
20 |
0.063 |
0.460 |
|
|
21 |
0.064 |
0.471 |
|
|
Rain |
19 |
0.122 |
0.553 |
|
20 |
0.114 |
0.544 |
|
|
21 |
0.118 |
0.549 |
|
|
Storm |
19 |
0.141 |
0.633 |
|
20 |
0.134 |
0.631 |
|
|
21 |
0.137 |
0.623 |
Table: Prediction results under different numbers of nodes
- Correct minor language errors.
We appreciate the reminder and regret the inconvenience. To improve the reading experience of the audience, the following modifications have been made to the complete text. The horizontal line shows the content that was eliminated, while the red symbolizes the text that was added.
Line 63: The development of soft measuring techniques has benefited from the introduction of several machine learning algorithms in recent years and have has performed well in the wastewater treatment sector.
Line 80: Instead of adjusting weights based on back-propagation of gradients, RVFL sets weights by Moore-Penrose generalized inverse. rather than adjust based on the results of each iteration.
Line 109: According to preliminary findings, Côté Cote et al. (1995) employed hybrid models comprising of a modified ASM1 model and FFNN models to accurately predict the concentrations of SS, COD, and NH4 in the effluent; DO in the bioreactor, and SS in the return sludge[23].
Line 132: 2.1 tTheory
Line 203: where Where Xj indicates the jth parameter of the best universe formed so far, lbj shows the lower bound of the jth variable, ubj is the upper bound of the jth variable, and r2, r3, r4 are random numbers in [0, 1].
Line 321: For the reason that only So, Sno, Snh, and Salk isare readily measurable through the laboratory or online instruments, the selected predictor variables are shown in Table 1.
Line 359: The effectiveness of the MVO-RVFL approach suggested in this study in predicting water quality, the original RVFL method, the conventional machine learning algorithm support vector regression algorithm(SVR), and Long short-term memory(LSTM) were chosen to compare it.
Line 410: In terms of model fit, RSME of the model with knowledge of the mechanism is 0.090, compared to the original MVO-RVFL model reduced by 32.8%.
Reviewer 2 Report
The authors have proposed a novel algorithm based on the random vector functional link network and Multi-Verse optimizer to forecast effluent quality. The paper is well written and organised. However, I would recommend major revision before it can be accepted in Sustainability journal. The following are my detailed comments to improve the manuscript.
1. Abstract: The sentence “Some…..in realtime” is not making any sense. Please revise.
2. I recommend rewriting abstract to show novelty and results of the study briefly.
3. I recommend citing this work 10.1007/s41101-021-00108-x to reinforce the introduction.
4. Line 94: Côté et al. (1995): Please follow same referencing style throughout the manuscript.
5. Please briefly explain the novelty of this manuscript in the last paragraph of introduction.
6. Section heading 2.1 : Theory
7. The quality of Figure 1 should be improved.
8. Please include a section on how the realtime data was obtained and how was its accuracy tested.
9. The authors should shorten the conclusions section.
10. The authors should make an effort to clean up the grammar and syntax in their manuscript. It will add to the readers’ enjoyment and appreciation for the work.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 3 Report
The subject of this paper is interesting. However, needs minor revision.
1. Include the more water treatment process using very recent references.
2. Establish the novelty of the work.
3. In figure 5: Include the captions inside the figure.
4. Check the caption of figure 6.
5. The authors should improve the MVO-RVFC model mechanism.
6. More typo errors in the manuscript; check it very carefully.
7. Rewrite the conclusion part.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
The authors have addressed all the comments. I recommentd acceptance of the paper for Sustainability.
