Optimizing Controlled Environmental Agriculture for Strawberry Cultivation Using RL-Informer Model
Round 1
Reviewer 1 Report
The manuscript that I have reviewed presents a good approach, is well written, with excellent structure and meets the conditions for publication. These are my recommendations for you to accept.
As a reviewer, here are my questions and recommendations:
1. The introduction provides a good overview of the importance of Controlled Environmental Agriculture (CEA) and the role of IoT technology in monitoring and predicting crop growth. However, it would be helpful to provide more specific information on the challenges faced in CEA, such as the limitations of traditional algorithms, the difficulties in integrating heterogeneous data, and the lack of real-time feedback. This would help readers understand the need for the proposed RL-Informer model.
2. The article mentions several studies that have used deep learning methods for crop growth prediction. It would be beneficial to provide more details on the methodologies and results of these studies, as well as their limitations. This would provide a better context for the proposed RL-Informer model and highlight its unique contributions.
3. The introduction briefly mentions that RL-Informer incorporates reinforcement learning to modify environmental variables. It would be helpful to provide more information on how this reinforcement learning process works and how it contributes to the overall prediction and feedback capabilities of the model.
4. The article states that RL-Informer is based on the Informer network, which is an improved network based on Transformer. It would be useful to provide a brief explanation of the Transformer model and its relevance to crop growth prediction. This would help readers understand the foundation on which RL-Informer is built.
5. The article mentions that RL-Informer focuses on predicting the growth of strawberries. It would be beneficial to explain why strawberries were chosen as the specific plant for this study and whether the model can be applied to other crops as well. Additionally, it would be helpful to provide some information on the specific environmental variables and characteristics of the plant that are considered in the prediction process.
6. The article mentions that RL-Informer provides feedback on modifying environmental variables to cater to diverse cultivation objectives. It would be valuable to provide examples of how this feedback mechanism works and how it can be utilized by farmers or growers to optimize their cultivation practices.
7. It would be beneficial to include a section on the methodology of the RL-Informer model, explaining the specific steps and algorithms involved in the prediction and feedback processes. This would provide readers with a clearer understanding of the technical aspects of the model.
8. Can you provide more information about the specific environmental variables collected in the experiment? It would be helpful to understand the range and variability of these variables.
9. How were the low-frequency and high-frequency data divided? What was the rationale behind this division?
10. The paper mentions the use of a central air conditioner and multiple wall-mounted air conditioners to control the temperature. Can you provide more details on how these devices were used and how the temperature was controlled?
11. It would be beneficial to provide more information about the training process of the Informer network. What was the size of the training and validation sets? Were any data preprocessing techniques applied?
12. The paper mentions the use of Q-learning for reinforcement learning. Can you provide more details on how the Q-learning algorithm was implemented and how the rewards were calculated?
13. It would be helpful to provide more information about the data collection process. How were the strawberry plants monitored and measured? Were there any challenges or limitations in collecting the data?
14. The paper mentions that the collected data was used to train the Informer model. Can you provide more details on the training process, such as the hyperparameters used and the loss function?
15. In the validation section, it would be useful to include a comparison of the results obtained from the RL-Informer model with those obtained
16. In Figure 7, it would be helpful to include the actual values of the predicted targets for each model, in addition to the visual representation. This would allow readers to better understand the performance of each model.
17. In Table 4, it would be useful to include the sample size for each evaluation metric. This would provide more context for the evaluation results.
18. In Figure 10, it would be helpful to include error bars or confidence intervals to indicate the variability in the biological parameters across the different groups. This would provide a better understanding of the statistical significance of the differences observed.
19. In Figure 11, it would be interesting to see the yield per plant or per unit area for both the control and RL-Informer guided groups. This would provide a more comprehensive comparison of the yield enhancement impac
Finally, it would be helpful to include a section on the results and evaluation of the RL-Informer model. This could include information on the accuracy of the predictions, the effectiveness of the feedback mechanism, and any comparisons with other existing models or methods.
Overall, the introduction provides a good overview of the topic and the proposed RL-Informer model. However, adding more details on the challenges, methodologies, and results would greatly enhance the clarity and comprehensiveness of the article.
No have comments.
Author Response
Our response is in the PDF file.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The article is devoted to the interesting and relatively promising method of Controlled Environmental Agriculture. The authors tried to describe a new method using predictions to optimize the process of growing plants, in this case, garden strawberries. I have several comments and questions about the work. I ask the authors to clarify:
- Part 2 of the article, Materials and Methods is insufficient.
- It is not entirely clear how the growing conditions were created and what the plants looked like. Photographs from experimental cultivation and arrangement in the cultivation laboratory must be documented.
- In particular, the description of measuring devices needs to be specified.
- How was the gradual ripening of the fruits manifested during the implementation of the research?
- In the work, including the Conclusion, the authors also mention the positive effect of the new method used on the fruit harvest, but nowhere is it described what the specific results of the harvest of individual strawberry varieties were. Was the yield evaluated only according to the number of fruits or was the weight of the fruits also evaluated? Photographs of the fruits must be documented and included.
- In the Conclusion, the authors state that the RL-Informer model is not only applicable to strawberry cultivation. It would be appropriate to state how they substantiate this claim. Do they have experience using this model for other crops? What are the characteristics of these crops? If experiments with other crops have not taken place, authors cannot make this claim in Conclusions.
- In the text in many cases, there is no space between the value (number) and the unit (dimension), e.g. row 146, and further in many cases in the text "350μmol/(m2 · s)", it should be: "350 μmol/(m2 · s)"
line 149 – the calculation relationship is not given and "The relative intensities of the light components" is not explained.
Figure 2 - the unit on the vertical axis is not shown.
Figure 6 – units (dimensions) on both axes are not shown.
- There are some formal flaws, typos, and errors in the English language in the work that need to be corrected.
- I recommend checking the format of text, figures, and tables according to the Template and Author Guidelines.
- There are some formal flaws, typos, and errors in the English language in the work that need to be corrected.
Author Response
Our response is in the PDF file.
Author Response File: Author Response.pdf
