Assessing Whole-Body Vibrations in an Agricultural Tractor Based on Selected Operational Parameters: A Machine Learning-Based Approach
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThe authors analysed whole-body vibrations of agricultural tractor operators using a machine learning approach. Data analysis was carried out using linear regression models. Three machine learning algorithms were used: Gradient Boosting Regressor, Support Vector Machine (SVM) Regressor and Multi-Layer Perceptron (MLP) Regressor. The results show that the Gradient Boosting Regressor is the most accurate model for the prediction of vibrations in the x- and y-axes, while the SVM Regressor performed best for the z-axis.
The article is particularly interesting and innovative, especially in relation to the aim of the journal. The potential implications in terms of operator comfort and safety are important.
The methodology is clearly described and well structured. The results are presented by the authors in detail and reflect the experimental hypotheses.
I would like to leave a number of suggestions that I consider essential to improve this manuscript.
1. Further detail the literature analysis, which, although present, is not exhaustive.
2. Make a comparison with the legal limits imposed by current legislation.
3. Please give more detail on the construction characteristics of the tractor tested.
4. Specify the driving system used and discuss its implications.
5. Based on my experience, I would suggest analysing the role of speed and tyre pressure more closely.
Author Response
Response to Reviewer 1
Thanks to the reviewer for a professional and constructive review! Please note that lines regarding the implemented changes in the manuscript refer to Track changes version with all markup.
Point: Further detail the literature analysis, which, although present, is not exhaustive.
Response: Thank you for your thoughtful feedback on the literature analysis in the manuscript. We sincerely appreciate your emphasis on ensuring a comprehensive review of existing work, as this strengthens the foundation of our study. In response to your comment, we have thoroughly revised the Introduction to expand the literature review. Specifically, additional references – we incorporated recent and foundational studies to better contextualize the relationship between Whole-Body Vibrations (WBV), operational parameters, and health risks in agricultural settings, and structured synthesis – to improve clarity and depth, we organized the literature into four subsections: WBV in Agricultural Tractors: Health Risks and Regulatory Context (highlighting gaps in current standards), Limitations of Existing Research (addressing methodological shortcomings), The Untapped Potential of Operational Parameters (emphasizing understudied variables), and Advancing WBV Prediction Through Machine Learning (bridging gaps with modern techniques). Please see lines 28-338.
Point: Make a comparison with the legal limits imposed by current legislation.
Response: We appreciate the reviewer’s suggestion. We have written the requested comparison, please see the lines 594-617.
Point: Please give more detail on the construction characteristics of the tractor tested.
Response: We appreciate the reviewer’s suggestion. We have written the requested construction characteristics of the tractor, please see the lines 376-377.
Point: Specify the driving system used and discuss its implications.
Response: We appreciate the reviewer’s suggestion. We have written the requested suggestion, please see the lines 366-375.
Point: Based on my experience, I would suggest analysing the role of speed and tyre pressure more closely.
Response: We appreciate the reviewer’s suggestion. We have written the requested suggestion, please see the lines 566-593.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsI have read the manuscript titled ‘Assessing Whole-body Vibrations in an Agricultural Tractor 2 Based on Selected Operational Parameters: A Machine Learn- 3 ing-Based Approach’ with interest. Here are the major concerns I observed with must be addressed before acceptance
1. The first sentence in the abstract must be completely changed to reflect the importance of the study. For instance, why is the study important. This must be coin out from the title, that is why is the study of tractor whol-body vibration important
2. Line 19 – 22 gave some important information ……… . The most 19 accurate machine learning model according to the R2 metric was the Gradient Boosting regressor 20 for the x-axis and the y-axis and for the z-axis the most accurate machine learning model was the 21 SVM regressor……… Please, state the r-square value
3. The model evaluation metrics used in this study are okay. But I expect a more robust one like normalized root mean square error (NRMSE) metric to be included
4. Please, which software was used for running the model…. Was any script written and run or a tool embedding in the software was used. Please, describe the methodology used in details
5. What version of the software was used
6. Please, categorically state the parameters that served as input and the output, particularly for the assessment of the tractor vibration
7. There are other machine learning models like ANN, ANFIS and so on. Why have the authors selected Gradient Boosting regressor; Support Vector Machine regressor; Multi-layer perception. Please, justify. Moreso, recently, the ensemble of some of the ML has been suggested to give better result, why is an ensemble of any of the ML models not considered in this study for improved prediction ?
8. Table 4; why is the result for train and validation together. The data used for training is different from validation and different from testing. So different results are expected. Please, the ratio of splitting is based on what for more clarity. Because, in this study authors claim 4/6 was used for training and validation together while 2/6 was used for testing. Honestly, training and validation can’t be together. You have to separate them and distinctly tell us the ratio of splitting vis-a-viz training: validation: testing
9. Include the unit of the RMSE and MAE in Table 4 or are the quantities measured dimensionless ?
10. Check the grammatical expressions again and very well
Comments on the Quality of English LanguageMajor English Revision Required
Author Response
Response to Reviewer 2
Thanks to the reviewer for a professional and constructive review! Please note that lines regarding the implemented changes in the manuscript refer to Track changes version with all markup.
Point: The first sentence in the abstract must be completely change to reflect the importance of the study. For instance, why is the study important. This must be coin out from the title, that is why is the study of tractor whole-body vibration important
Response: We appreciate the reviewer’s suggestion. We changed the first sentence in the abstract. Please see lines 10-11.
Point: Line 19 – 22 gave some important information ……… . The most 19 accurate machine learning model according to the R2 metric was the Gradient Boosting regressor 20 for the x-axis and the y-axis and for the z-axis the most accurate machine learning modelwas the 21 SVM regressor……… Please, state the r-squarevalue
Response: We appreciate the reviewer’s comment. We state the R2 value in the abstract. Please see lines 20-21.
Point: The model evaluation metrics used in this study are okay. But I expect a more robust one like normalized root mean square error(NRMSE) metric to be included
Response: We appreciate the reviewer’s suggestion. We included the NRMSE metric. Please see line 536, the formula for NRMSE and the updated Table 6 (line 620), Table 7 (line 621) and Table 8 (line 622).
Point: Please, which software was used for running the model…. Was any script written and run or a tool embedding in the software was used. Please, describe the methodology used in details
Response: We appreciate the reviewer’s suggestion. We described our software setup in a separate paragraph. Please see lines 418-422.
Point: What version of the software was used
Response: We appreciate the reviewer’s suggestion. We described our software setup in a separate paragraph (alongside the versions of the software). Please see lines 418-422.
Point: Please, categorically state the parameters that served as input and the output, particularly for the assessment of the tractor vibration
Response: We appreciate the reviewer’s suggestion. We explicitly stated the parameters that served as the input and the output. Please see lines 413-416.
Point: There are other machine learning models like ANN, ANFIS and so on. Why have the authors selected Gradient Boosting regressor; Support Vector Machine regressor; Multi-layer perception. Please, justify. Moreso, recently, the ensemble of some of the ML has been suggested to give better result, why is an ensemble of any of the ML models not considered in this study for improved prediction ?
Response: We appreciate the reviewer’s suggestion regarding the justification of model selection. We have elaborated on this point in lines 470-475. We would also like to clarify that the multi-layer perceptron (MLP) used in our study is a type of Artificial Neural Network (ANN), meaning that ANN-based approaches have already been incorporated in our analysis. Furthermore, the Gradient Boosting Regressor (GBR) is an ensemble-based method that combines multiple decision trees, meaning ensemble learning has already been applied in our work.
Point: Table 4; why is the result for train and validation together. The data used for training is different from validation and different from testing. So different results are expected. Please, the ratio of splitting is based on what for more clarity. Because, in this study authors claim 4/6 was used for training and validation together while 2/6 was used for testing. Honestly, training and validation can’t be together. You have to separate them and distinctly tell us the ratio of splitting vis-a-viz training: validation:testing
Response: We appreciate the reviewer’s comment regarding the splitting of the training and validation sets. As described in Section 2.2.3 (lines 507-514), we performed a grid search with 10-fold cross-validation on the train & validation subset of the data, where each fold consisted of 90% training and 10% validation data. This approach allowed us to optimize hyperparameters robustly and re-train the model on the entire train & validation set on the best hyperparameters found (which are reported in Table 4 for each model). The final evaluation in Table 6, Table 7 and Table 8 reflects the performance of the model on the combined train & validation set, while the test set results are reported separately. We have also clarified the data splitting ratio in the manuscript, noting that 80% of the total data was used for training and validation, and 20% for testing (see lines 464-466).
Point: Include the unit of the RMSE and MAE in Table 4 or are the quantities measured dimension less?
Response: We appreciate the reviewer’s comment. We added this in Table 6, Table 7 and Table 8.
Point: Check the grammatical expressions again and very well
Response: We appreciate the reviewer’s comment. We read the entire paper again after all of the revisions and improved the sentence construction.
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments and Suggestions for AuthorsThe evaluated manuscript aims to predict whole-body vibrations experienced by a tractor operator inside the cabin of an agricultural tractor along three axes (x, y, and z). The input features include agrotechnical surface type, movement speed, and tire air pressure, while the target variable is vibration along the respective axes.
Overall, the manuscript is well-written, scientifically valuable, and easy to comprehend. Assessing tractor vibrations is crucial for improving ergonomic conditions in agricultural operations. The application of machine learning methods in this study effectively predicts whole-body vibrations for all axes, demonstrating promising results.
After a general assessment, I have the following comments and concerns:
1. Line 110: Remove the extraneous "W" at the end of the line.
2. Vibration Measurement Device: Provide more detailed specifications of the vibration measurement device, including its measurement range and accuracy, in the Materials and Methods section.
3. Clarify the designation of the x, y, and z axes on the tractor. A graphical representation of the coordinate system on the tractor would enhance clarity. Have orthogonal axes been used?
4. Clearly indicate the units of measurement, particularly for vibration, in Table 3 and Figure 2.
5. Table 3 Percentiles: The meaning of the 25%, 50%, and 75% values in Table 3 is unclear. An explanation should be provided.
6. Including the standard error of the mean in Table 3 would facilitate a more meaningful comparison of vibrations across the x, y, and z axes.
7. The font size in the charts of Figure 3 is too small. Consider increasing it for better readability.
8. The discussion is relatively weak. A more detailed and in-depth analysis of the results should be provided in the Results and Discussion section.
Overall, the study presents valuable findings, but addressing the above points will enhance the manuscript’s clarity, accuracy, and scientific rigor.
Comments on the Quality of English LanguageEnough to understand, but minor revision is needed.
Author Response
Response to Reviewer 3
Thanks to the reviewer for a professional and constructive review! Please note that lines regarding the implemented changes in the manuscript refer to Track changes version with all markup.
Point: Line 110: Remove the extraneous "W" at the end of the line.
Response: We appreciate the reviewer’s comment. It was removed; take a look at line 336.
Point: Vibration Measurement Device: Provide more detailed specifications of the vibration measurement device, including its measurement range and accuracy, in the Materials and Methods section.
Response: We appreciate the reviewer’s comment. We have written the requested specifications, please see the lines 384-385.
Point: Clarify the designation of the x, y, and z axes on the tractor. Agraphical representation of the coordinate system on the tractorwould enhance clarity. Have orthogonal axes been used?
Response: We appreciate the reviewer’s comment. We have inserted the requested figure, please see the lines 398-400.
Point: Clearly indicate the units of measurement, particularly for vibration, in Table 3 and Figure 2.
Response: We appreciate the reviewer’s comment. We have indicated the units of measurement clearly in Table 3 (now Table5) (line 564) and in Figure 2 (now Figure 3), 4 and 5. We also removed a column (N) from Table 3 (now Table5); we noted that all axes had 648 data samples associated with them (line 552).
Point: Table 3 Percentiles: The meaning of the 25%, 50%, and 75%values in Table 3 is unclear. An explanation should be provided.
Response: We appreciate the reviewer’s comment. Please see lines 552-562.
Point: Including the standard error of the mean in Table 3 would facilitate a more meaningful comparison of vibrations across the x, y, and z axes.
Response: We appreciate the reviewer’s comment. We added the standard error of the mean (SEM) in Table 3 (now Table 5 – line 564) as a column. We also added an explanation of what the SEM column stands for on lines 562-563.
Point: The font size in the charts of Figure 3 is too small. Consider increasing it for better readability.
Response: We appreciate the reviewer’s comment. We increased the font size in Figure 3 (now Figure 6) as suggested by the reviewer. Please see line 680.
Point: The discussion is relatively weak. A more detailed and in-depth analysis of the results should be provided in the Results and Discussion section.
Response: We appreciate the reviewer’s suggestion. We expanded upon the Results and Discussion section and revised some parts. Please see lines 551-689.
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
Comments and Suggestions for AuthorsI agree with the responses
Reviewer 3 Report
Comments and Suggestions for AuthorsI have carefully examined the revised version of your manuscript, reference number agriengineering-3480706.
Thank you for addressing my previous comments with thorough and thoughtful responses. I am pleased to see that all my concerns have been satisfactorily addressed. Your revisions and replies demonstrate a clear understanding of the issues raised, and the changes made have improved the overall quality of the manuscript.
