Identification Model of Soil Physical State Using the Takagi–Sugeno Fuzzy Neural Network
Round 1
Reviewer 1 Report
The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. However, several issues have observed into presented work such as:
-Related Work is missing as well as research gaps discussion, as in what gaps lead you to propose another approach?
-The contribution of the proposed study is not clear is ti more on data clustering or data classification or both? please provide list of study contributions at the end of related work section.
-The used dataset is not defined well and i do not know how far is reliable to suit data science processes. A new section is needed to describe the dataset.
-A reasonable justification should be made about why such algorithms are used for prediction and clustering purposes. Why do authors think they are appropriate for such application. What is their main advantage over other methods ?
-Is there any where the source code of paper like GitHub?
-The format of the paper is need to be revised and remove the extra sections.
-Data handling techniques training, validation, and testing datasets need to discuss in detail. Also the hyper-parameters settings should be highlighted into table form.
-Comparative study of the result finding with the proposed method with 3 to 4 existing methods though included graphs author can include result table with the parameters.
-Most of evaluation section was presented based on only accuracy, more accurate metrics need to be highlighted to prove the efficiency of the proposed method such as Recall, F-Score, and Precision.
-The conclusion part should include information about the obtained experimental values.
-Please explain more clearly the limitation and future work.
-Authors mentioned "This indicates that the T–S fuzzy neural network model based on ploughing resistance and working parameters can accurately identify the physical state of paddy soil in real-time during ploughing" my question how much time is required to get result by proposed model?
Author Response
Response to Reviewer 1 Comments
Thank you very much for your valuable comments.
Explanation: Sections need to be explained were provided in the resubmitted manuscript through comments. Please see the attachment.
Point 1: Related Work is missing as well as research gaps discussion, as in what gaps lead you to propose another approach?
Response 1: The innovation of this manuscript is the first time using fuzzy neural network to construct a model to prediction soil physical state. This prediction is very necessary for the adjustment of operating parameters in plowing. I am very sorry for the lack of content, and made further explanation in the article. (Section 1)
Point 2: The contribution of the proposed study is not clear is it more on data clustering or data classification or both? please provide list of study contributions at the end of related work section.
Response 2: The clustering algorithms used in this studyr include FCM and fuzzy c-means method based on subtractive clustering.The role of the algorithm is explained in the manuscript.(Table 1, Section 3.4)
Point 3: The used dataset is not defined well and I do not know how far is reliable to suit data science processes. A new section is needed to describe the dataset.
Response 3: I added a description of the dataset to the manuscript again. (Section 3.5)
Point 4: A reasonable justification should be made about why such algorithms are used for prediction and clustering purposes. Why do authors think they are appropriate for such application. What is their main advantage over other methods ?
Response 4: I added section about a reasonable explanation of using the algorithm and their main advantage over other methods. (Section 3.4)
Point 5: Is there any where the source code of paper like GitHub?
Response 5: Sorry, I'm not too clear about that. Because I used Matlab toolbox and program written by myself to complete this research
Point 6: The format of the paper is need to be revised and remove the extra sections.
Response 6: Thanks for your reminder I submitted the manuscript according to the format required by the journal, and the extra sections was automatically added by the system after I submitted the manuscript.
Point 7: Data handling techniques training, validation, and testing datasets need to discuss in detail. Also the hyper-parameters settings should be highlighted into table form.
Response 7: I added more content to discuss the training, validation, and testing datasets in detail. Meanmmile, I provided the hyper-parameters settings value ,not in the table form, taking into account the aesthetics of the manuscriot layout. (Section 3.5)
Point 8: Comparative study of the result finding with the proposed method with 3 to 4 existing methods though included graphs author can include result table with the parameters.
Response 8: Thank you very much for your valuable comments again. In this paper, the fuzzy c-means clustering method based on subtractive clustering is analyzed with FCM and genetic algorithm, but the comparative parameter values are not provided, because the advantages and disadvantages of different algorithms in identifying network parameters have been clarified in relevant research, So I don't think we need to make further comparison. However, if the editor thinks this is a very important content, I will make further modifications and hope to communicate with you further. because the advantages and disadvantages of different algorithms in identifying network parameters have been clarified in relevant studies(references was provided), so I do not think that further comparison is necessary. However, if the reviewer thinks this is a very important content, I will make further modifications, and hope to further communicate with you.( Section 3.4 and Section 3.5)
Point 9: Most of evaluation section was presented based on only accuracy, more accurate metrics need to be highlighted to prove the efficiency of the proposed method such as Recall, F-Score, and Precision.
Response 9: I provided more metrics. (Section 3.5)
Point 10: The conclusion part should include information about the obtained experimental values.
Response 10: I added the the obtained experimental values in the conclusion part. (Section 4 (2)(3))
Point 11: Please explain more clearly the limitation and future work.
Response 11: The section of the limitation and future work have been provided.(Section 3.4 and Section 4 (4))
Point 12: Authors mentioned "This indicates that the T–S fuzzy neural network model based on ploughing resistance and working parameters can accurately identify the physical state of paddy soil in real-time during ploughing" my question how much time is required to get result by proposed model?
Response 12: The results can be obtained after 0.5s. The existence of this delay is acceptable, because according to the operating velocity, the forward distance in the delay time is less than the distance between sampling points. (Section 3.5)
Author Response File: 
Author Response.docx
Reviewer 2 Report
The current paper suggested that the working parameters and soil physical parameters of ploughing were determined using a designed electric suspension platform and soil instrument, here the Takagi–Sugeno (T–S) fuzzy neural network classifier was constructed using traction resistance, operating velocity, and ploughing depth as inputs to indirectly identify the soil physical state. The theory is validated using simulations.
Comments to authors:
- I think that is after reference [29] should be eliminated since is part of the template
- Please add more details of how the theory from the first sections is applied in the results section.
- Define how the parameters of the Takagi-Sugeno fuzzy neural network controller were obtained
- The mathematical model of the process cand be added
- In the current paper the authors talk about optimization, why no optimization function is specified.
- The authors can add the steps of implementing the algorithms. The theoretical part can be better detailed. The steps will be in the benefit of the readers, maybe they’ll help the readers to implement the proposed algorithm.
- Add the measurement units labels for abscissa and ordinate for all the figures from the paper.
- The state of the art it is very poor regarding representative papers, maybe the author could add the following publications:
o Hybrid Data-Driven Fuzzy Active Disturbance Rejection Control for Tower Crane Systems, European Journal of Control, vol. 58, pp. 373-387-11, 2021.
o Enhanced P-type Control: Indirect Adaptive Learning from Set-point Updates, IEEE Transactions on Automatic Control, DOI: 10.1109/TAC.2022.3154347, 2022.
- Add the both the advantages and the disadvantages of the proposed method, in the current version of the paper only the advantages are presented.
Author Response
Response to Reviewer 2 Comments
Thank you very much for your valuable comments.
Explanation: Sections need to be explained were provided in the resubmitted manuscript through comments. Please see the attachment.
Point 1: I think that is after reference [29] should be eliminated since is part of the template
Response 1: Thanks for your reminder. I submitted the manuscript according to the format required by the journal, and the extra sections was automatically added by the system after I submitted the manuscript.
Point 2: Please add more details of how the theory from the first sections is applied in the results section.
Response 2: Added more detailed explanation. (Section 3)
Point 3: Define how the parameters of the Takagi-Sugeno fuzzy neural network controller were obtained.
Response 3: Added the process and results of obtaining controller parameters.(Section 2.4.1 and 3.5)
Point 4: The mathematical model of the process can be added.
Response 4: Added the mathematical model of the process.(Section 2.3.1 and 2.3.2)
Point 5: In the current paper the authors talk about optimization, why no optimization function is specified.
Response 5: Added the optimization function. (Section 2.4.2)
Point 6: The authors can add the steps of implementing the algorithms. The theoretical part can be better detailed. The steps will be in the benefit of the readers, maybe they’ll help the readers to implement the proposed algorithm.
Response 6: Added the steps of implementing the algorithms.(Section 2.4.1 and 2.4.2)
Point 7: Add the measurement units labels for abscissa and ordinate for all the figures from the paper.
Response 7: Added the measurement units labels for abscissa and ordinate for all the figures from the paper.(Section 3.2 (Figure 5.6) and Section 3.3 (Figure 7))
Point 8: The state of the art it is very poor regarding representative papers, maybe the author could add the following publications:
o Hybrid Data-Driven Fuzzy Active Disturbance Rejection Control for Tower Crane Systems, European Journal of Control, vol. 58, pp. 373-387-11, 2021.
o Enhanced P-type Control: Indirect Adaptive Learning from Set-point Updates, IEEE Transactions on Automatic Control, DOI: 10.1109/TAC.2022.3154347, 2022.
Response 8: Thank you very much for your valuable comments again. After checking the publications provided in comments and suggestions, I will share my views with you. The innovation of this study is the first time using fuzzy neural network to construct a model to prediction soil physical state. In this study, the role of the controller is to divide soil conditions into three categories, so it is also called classifier in this manuscript. The same as the controller is that the parameter setting of the classifier is based on data, which is consistent with the viewpoint in the publications provided in comments and suggestions. The purpose of placing publications at [17,18] is to illustrate that although the model in this study is to identify the soil physical state, the classifier can provide a basis for the adjustment of operating parameters during ploughing, because this is a classifier based on data-driven recognition.(Section 1 ,Reference[17,18])
Point 9: Add the both the advantages and the disadvantages of the proposed method, in the current version of the paper only the advantages are presented.
Response 9: Added the both the advantages and the disadvantages of the proposed method. (Section 3.5)
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
- - Comment 2 “The contribution of the proposed study is not clear is ti more on data clustering or data classification or both? Please provide list of study contributions at the end of related work section.” has not addressed by authors.
- - Authors mentioned “According to the results of the ploughing experiment in chapter 2.6 and the division of soil physical state in chapter 3.3, 720 datasets of traction resistance, working parameters (operating speed and ploughing depth)” which chapters authors refer to??.
- - Comment 3 is not addressed well by authors where no new section has been added. However, the details of collected dataset are overlapped with proposed model. Furthermore, still I have some worries about reliability of the data especially with few collected samples. Authors have to present their results in order to show there is no overfitting or underfitting issue.
- - Comment 8 has not addressed by author. However, authors claimed that there is no need to compare with state of the art methods since they already covered all merits and demerits of some of existing studies where this no the whole story. In computer science and data science aspects, since you collected a new data or propose a new model in order to reproduce your study and make benefits for other scholars you have to run your model on different datasets as well as your collected dataset on different models in order to generalize your work and show the efficiency of produced study. Therefore, I insist to compare your work with 2 to 3 recent works based on used evaluation metrics.
Author Response
Response to Reviewer 1 Comments
Thank you very much for your valuable comments.
First of all, thank you for your criticism and suggestions. I learned a lot about the major of article writing
Explanation: Sections need to be explained were provided in the resubmitted manuscript through comments. Please see the attachment.
Point 1: Comment 2 “The contribution of the proposed study is not clear is it more on data clustering or data classification or both? Please provide list of study contributions at the end of related work section.” has not addressed by authors.
Response 1: Thank you very much for the valuable comments I had provided the clustering methods used in the manuscript and the role of these methods in the study. You can see this part from table 1.If I have a wrong understanding of the comment, I hope you can give me similar examples and we can communicate further.
Point 2:Authors mentioned “According to the results of the ploughing experiment in chapter 2.6 and the division of soil physical state in chapter 3.3, 720 datasets of traction resistance, working parameters (operating speed and ploughing depth)” which chapters authors refer to ??
Response 2: I'm sorry for the bad impression that my writing has brought you. The input and output of the model are obtained from the ploughing test in section 2.6, and the input of the model is the values of traction resistance, operating speed and ploughing depth obtained in the test. Through the clustering method in section 3.3, the soil parameters are divided into three states of ‘soft’, ‘zero’and ‘hard’, and the value 1, 2 and 3 are given as the output of the model respectively.
Point 3: Comment 3 is not addressed well by authors where no new section has been added. However, the details of collected dataset are overlapped with proposed model. Furthermore, still I have some worries about reliability of the data especially with few collected samples. Authors have to present their results in order to show there is no overfitting or underfitting issue.
Response 3: On the basis of adding the comparison method, it is finally determined that the fuzzy c-means clustering based on subtractive clustering is used to identify the structural parameters of the antecedent network, and the hybrid learning algorithm is used to optimize the parameters of the consequent network, which is more suitable for this study. After determining the method, a figure( Figure 9) of the curve of RMSE changing with the number of iterations is provided to show that there is no overfitting or underfitting issue.( Section 3.5.4 and Figure 9)
Point 4: Comment 8 has not addressed by author. However, authors claimed that there is no need to compare with state of the art methods since they already covered all merits and demerits of some of existing studies where this no the whole story. In computer science and data science aspects, since you collected a new data or propose a new model in order to reproduce your study and make benefits for other scholars you have to run your model on different datasets as well as your collected dataset on different models in order to generalize your work and show the efficiency of produced study. Therefore, I insist to compare your work with 2 to 3 recent works based on used evaluation metrics.
Response 4: Thank you very much for your valuable comments again. From the comments, I learned about the academic rigor, which further increased my academic literacy. The change process of the objective function of the fuzzy C-means clustering and the fuzzy C-means clustering based on subtractive clustering was added in the manuscript adds when identifying the network structure parameters of the antecedent, as shown in the Figure 8. At the same time, Recall, F-Score, and Precision are selected as accurate metrics, and the results of hybrid learning algorithm, gradient descent method and genetic algorithm in optimizing parameters are compared, as shown in the Table 2.(Section 3.5)
Author Response File: Author Response.docx
Reviewer 2 Report
The authors seriously improved the paper since last revision. From my point of view the paper can be accepted as contribution in Agriculture journal.
Author Response
Thanks for the comments, and I learned a lot from your comments. Thank you again.
Round 3
Reviewer 1 Report
Authors have covered all issues into present version. Now the paper can be accepted.
Author Response
Response to Academic Editor's Comments
Thank you very much for your valuable comments.
First of all, thank you for your criticism and suggestions. I learned a lot about the major of article writing
Explanation: Sections need to be explained were provided in the resubmitted manuscript through comments. Please see the attachment.
Point 1: Comment 2 was “The contribution of the proposed study is not clear. Is it more on data clustering or data classification or both? Please provide list of study contributions at the end of related work section.” The authors’ response was “The clustering algorithms used in this study include FCM and fuzzy c-means method based on subtractive clustering. The role of the algorithm is explained in the manuscript.” I agree with the reviewer—the authors have not explicitly answered what the study’s contribution is. The reviewer’s comment had nothing to do with clustering algorithms; rather, it had to do with the import or impact of the manuscript. To underscore this, in his comment 1, the reviewer stated that related work was missing in the literature reviewer, and the research gaps that led to approach described in the manuscript had not been identified. The authors responded “The innovation of this manuscript is the first time using fuzzy neural network to construct a model to prediction soil physical state.” But this is clearly not so. In addition to literature cited by the authors, the following, easily-searched articles used a “fuzzy neural network to construct a model to prediction soil physical state”:
Soil shear strength prediction using intelligent systems: artificial neural networks and an adaptive
neuro-fuzzy inference system (2012);
Application of Fuzzy-Neural Network in Classification of Soils using Ground-penetrating Radar
Imagery (2013);
Estimation of saturated hydraulic conductivity using fuzzy neural network in a semi-arid basin scale
for murum soils of India (2018);
Soil temperature modeling at different depths using neuro-fuzzy, neural network, and genetic
programming techniques (2017); and
A fuzzy‐neural network method for modeling uncertainties in soil‐structure interaction problems
(2003).
The reviewers should therefore respond better to Reviewer #2’s Comment 2.
Response 1: I ' m deeply sorry for my unclear statement, which gives you a bad reading experience. I restated the innovation of the article as follows :
In the ploughing process, identifying the physical state of soil is of great significance for the adjustment of operation parameters. Although the article you provide uses ' fuzzy neural networks to build models for predicting soil physical state ', the problems we solve are different. This manuscript is more suitable for identifying the physical state of soil in the ploughing process and is biased towards perception application in tractor operation, which is different from the significance of the article you provide. Few studies have been conducted on the use of a soil condition identification model during the ploughing process.In this paper, the fuzzy neural network is used to establish the soil physical state recognition model, which contributes to the adjustment of operating parameters in the process of plowing.
and revised this part in the manuscript.( Section 3.5.4)
Point 2: Comment 8 was “Comparative study of the result finding with the proposed method with 3 to 4 existing methods though included graphs author can include result table with the parameters.” The author’s response was “In this paper, the fuzzy c-means clustering method based on subtractive clustering is analyzed with FCM and genetic algorithm, but the comparative parameter values are not provided, because the advantages and disadvantages of different algorithms in identifying network parameters have been clarified in relevant research, So I don't think we need to make further 2 comparison. However, if the editor thinks this is a very important content, I will make further modifications and hope to communicate with you further. Because the advantages and disadvantages of different algorithms in identifying network parameters have been clarified in relevant studies (references was provided), so I do not think that further comparison is necessary. However, if the reviewer thinks this is a very important content, I will make further modifications, and hope to further communicate with you.” The reviewer’s response to this was “Comment 8 has not (been) addressed by author. Authors claim that there is no need to compare with state of the art methods since they already covered all merits and demerits of some of existing studies, whereas this not the whole story. In computer science and data science, since you collected new data or propose a new model, you should run your model on different datasets, and different models on your data set, in order to generalize your work by comparing and contrasting your study results to those of two to three recent works, using evaluation metrics.” In this case, I am going to agree with the reviewer, and ask the authors to comply to the extent possible with the italicized text.
Response 2: Thank you very much for your valuable comments again.The manuscript you saw is the result after the revision of my first review, not the latest manuscript, and I do not know why such a result is produced. I 've made modifications in the latest manuscript.
The change process of the objective function of the fuzzy C-means clustering and the fuzzy C-means clustering based on subtractive clustering was added in the manuscript adds when identifying the network structure parameters of the antecedent, as shown in the Figure 8. At the same time, Recall, F-Score, and Precision are selected as accurate metrics, and the results of hybrid learning algorithm, gradient descent method and genetic algorithm in optimizing parameters are compared, as shown in the Table 2.(Section 3.5.4)
Point 3: Finally, there remain several typos and formatting issues that need to be addressed. The most
important among these are the graphs. They all have incomplete captions. The caption should
sufficiently explain the graph such that readers can understand it without having to refer to the text
body. That is, graphs should “stand on their own”. None of the graph captions currently do this for
any of the graphs. Also, graph axes should include SI units of measurement. For example, “%” is
used in Fig. 5 for volumetric soil moisture content. But % is not an SI unit. Figure 7 has no units at all.
Terms don’t always agree either. For example, the text in section 3.5 states that soils that had “soft,
solid and hard conditions” but the graph has “soft,” “hard,” and “zero” in the legend.
Some of the typos include “onstruted” in section 2.3, “The four layer” in section 2.3.1, and “Figure”
with no subsequent number in section 3.2. The manuscript needs to be reviewed and revised to
correct these minor errors. Granted they are minor, but they make the manuscript harder to read and
interpret.
In figure 7 in the caption I would mention the "relative units" of the axes.
Response 3: Thank you again for your comments on the details in the manuscript. I have modified the detailed errors you mentioned, and the modified parts are marked in the manuscript.
Author Response File: Author Response.docx
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.
