An Improved Multi-Objective Optimization Decision Method Using NSGA-III for a Bivariate Precision Fertilizer Applicator
Round 1
Reviewer 1 Report
This paper is called “An improved multi-objective optimization decision method using NSGA-III for bivariate precision fertilizer applicator “.
Some specific comments are as follows:
-Although you stated that fertilizer properties are important in the material method section, you used only one fertilizer. For example, if you use ammonium sulfate fertilizer, I think the system will not give the same successful results. Therefore, it would be more appropriate to take fertilizers with different properties as factors in this study.
-Could you explain which components you used to obtain 345 data in your study? What does each group represent? Can you elaborate?
-The number of data used in training, testing, and validation should be written.
-It would be more appropriate to move the paragraph starting with line 323 to the Results and Discussion section.
-What is the reason why the Fertilization Rate 5 and 7 in Table 5 are very high?
-The Results and Discussion section should be discussed.
-Figure numbers should be checked.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
In this study, the authors proposed a fertilization decision method in order to find the optimal combination of the rotational speed of the fertilizer discharging shaft and the opening length of the feeding inlet of a developed bivariate granular fertilizer applicator. The accuracy, uniformity, adjustment time, and breakage rate are considered as optimization objectives. To collect the fertilization data, a bench experiment is conducted. Using rotational speed and opening length as input parameters and fertilization rate as output, four machine learning algorithms were incorporated to build the prediction model of fertilization rate based on two data segments, A and B. Accordingly, four objective functions are obtained for four optimization objectives. NSGA-III algorithm was adopted to find the optimal fertilization decision based on the considered objectives. Hypervolume indicator is used to evaluate the convergence and diversity of the observed solution sets. NSGA-III is compared to GA and MOEA-D-DE algorithms proposed in the literature. Assessment criteria for each objective are proposed and the results are presented. Using machine learning algorithms final prediction model is successfully obtained. The assessment criteria are measured and the comparative results of the considered algorithms are presented graphically. The results indicated the reduction in assessment criteria of four objective functions: average relative error (accuracy), coefficient of variation (uniformity), adjustment time, and breakage rate. As confirmed, the presented method showed significant improvements in accuracy and adjustment time compared to GA and MOEA-D-DE, whereas uniformity and breakage rate are within approriate ranges with slight improvements compared to GA and MOEA-D-DE.
In my opinion, the strengths of the proposed manuscript are:
Brief state-of-the-art review of fertilizer applicators, systems, and parameters as well as multi-objective optimization algorithms
Contributions regarding fertilization data collection, segmentation, their rationality, and use of machine learning algorithms to build fertilization rate prediction model
Concise presentation of objective functions and assessment criteria for evaluating fertilization performance
Correctly formulated and presented NSGA-III algorithm used for optimizing the fertilization decision
Detailed representation of experimental results regarding the performance of fertilization rate prediction model and the multi-objective optimization using NGSA-III algorithm
The weaknesses of the manuscript are:
Please check technical details such as figure labels:
Line 301 – Should be „Figure 4“
Line 473 – Should be „Figure 5“
The concluding remarks could be enriched. Authors could provide more information on results obtained by ML algorithms regarding the prediction model, as well as results for all four objectives using NSGA-III. Also, directions for future research regarding optimization in fertilization applications should also be considered.
Author Response
Please see the attachment.
Author Response File: 
Author Response.docx
Reviewer 3 Report
This paper studies the optimization of fertilization decision of bivariable variable fertilizer machine . The data collected from the bench test are processed by outliers and segmentation. The prediction model of fertilization rate is obtained by machine learning algorithm. The fertilization performance model aiming at accuracy, uniformity, adjustment time and breakage rate was established and solved by NSGA-III algorithm to obtain the optimal fertilization decision. This paper can provide reference for the research of variable fertilizer applicator. However, there are some minor problems, which should be improved before publication. There are given below:
(1). The utility of the fertilizer rate prediction model is not clearly described in the abstract.
(2). In section 2.3.2, the author mentions the concept of average growth rate of fertilization rate, but the description of formula (2) is changed to “The average growth rate of opening length”.
In addition, the description of the object and calculation steps of variance calculation in the segmentation method is not specific enough.
(3). In Algorithm 1 on Pg.10, it seems that line 11 needs to be indented.
(4). In Section 2.6, the two methods mentioned by the author for comparison with the method in this paper lack references.
(5). The formula format is not standard enough, and some formula numbers have alignment errors
(6). Conclusions needs more in it, as it's more of an afterthought. The authors are suggested to highlight important findings and include afterthought of this work.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
Can you explain what components you used to obtain 345 data in your study?
