Feature Extraction of Anomaly Electricity Usage Behavior in Residence Using Autoencoder

Round 1

Reviewer 1 Report

Major recommendations

1. The Introduction Section: It is too weak. There are only 6 references. The current state of the research field should be carefully reviewed and key publications cited.
2. The main aim of the study has to be at the end of the Introduction Section. If the aim is unknown, that it is difficult to understand the article.
3. The conclusion Section should be rewrite in a better way.

 

Minor recommendations

1. Line 16: “As global warming steadily worsens ….”, please rephrase.
2. Lines 28-29: “ … this study extracts high-level features of the annual electricity consumption of five actual residential users ….” In my opinion, the authors should substitute the term “extract” by more suitable one.
3. Line 31, keywords: “energy saving and carbon reduction” in my opinion, it is better to use two separate expressions: “energy saving; carbon reduction”.
4. Figures 2 and 3: The authors use the following terms: None, 2; None, 96; None, 64, etc. They must be described.
5. Line 281: It is better to place a formula in the Research Methods Section.
6. Figures 5-9 are too small. It is difficult to read them.

Author Response

Thank you very much for the reviewer for the comments and suggestions. We have revised the manuscript according to the comments and now explain to the reviewer as follows:

Question 1: The Introduction Section: It is too weak. There are only 6 references. The current state of the research field should be carefully reviewed and key publications cited.

Response 1: Thank you for this valuable suggestion. We have reviewed the recent studies in this research field and citied the related papers in the introduction of this revised manuscript. Additionally, we also cited other related papers about the anomaly detection of power consumption in the second section to provide a complete background review.

Question 2: The main aim of the study has to be at the end of the Introduction Section. If the aim is unknown, that it is difficult to understand the article.

Response 2: Thank you for this valuable suggestion. We have modified the description of the Introduction section to highlight the explanation of the main aim of this study.

Question 3: The conclusion Section should be rewrite in a better way.

Response 3: The conclusion section has been modified and rewritten to explain the results and future works in this study clearly.

Question 4: Line 16: “As global warming steadily worsens ….”, please rephrase.

Response 4: We have rephrased this description in the revised manuscript.

Question 5: Lines 28-29: “ … this study extracts high-level features of the annual electricity consumption of five actual residential users ….” In my opinion, the authors should substitute the term “extract” by more suitable one.

Response 5: To express the meaning of this sentence clearly, we rewrote it and substituted the term “extract” in the revised manuscript. Thank you for this suggestion.

Question 6: Line 31, keywords: “energy saving and carbon reduction” in my opinion, it is better to use two separate expressions: “energy saving; carbon reduction”..

Response 6: Thank you for this valuable suggestion. We have modified the keywords in the revised manuscript.

Question 7: Figures 2 and 3: The authors use the following terms: None, 2; None, 96; None, 64, etc. They must be described.

Response 7: Thank you for this valuable suggestion. The intention of the first and the second dimensions of input and out shapes are the batch size and input size, respectively. And the term “None” in the first dimension means that the batch size depends on how many samples we give for training. We also added explanations about these terms in the revised manuscript. 

Question 8: Line 281: It is better to place a formula in the Research Methods Section.

Response 8: The formula has been moved to the Research Methods section.

Question 9: Figures 5-9 are too small. It is difficult to read them.

Response 9: Thank you for this valuable suggestion. We have repainted the figures and modified the presentation modes of these figures to be easy to read.

Author Response File: Author Response.pdf

Reviewer 2 Report

This study adopts an autoencoder, which is a deep learning technology, to extract the high-level electricity consumption information of residential customers in an attempt to improve the anomaly detection performance of the model.

 

This topic is very interesting and this paper is also well written. To enhance the quality of the manuscript, please consider including the comments below.

 

1. Some figures shown in this paper are not very clear. Please try to improve the quality of all figures in the revised version.

 

2. In introduction part, the contributions of the paper should be rewritten clearly.

 

3. The computational complexity of deep-learning-based autoencoder for anomaly behavior detection should be analyzed.

 

4. The proposed deep-learning-based autoencoder should be further compared with more advanced deep learning approaches.

 

5. The author is invited to perform a thorough proofread of their manuscript, as I can still spot some spelling/grammar mistakes in the paper.

 

6. The literature review about anomaly data detection are not sufficient in the current version of this paper. For example, the following manuscript titled as “Robustness of Short-term Wind Power Forecasting against False Data Injection Attacks” is suggested to be included.

Author Response

Thank you very much for the reviewer for the comments and suggestions. We have revised the manuscript according to the comments and now explain to the reviewer as follows:

Question 1: Some figures shown in this paper are not very clear. Please try to improve the quality of all figures in the revised version.

Response 1: Thank you for this valuable suggestion. We have repainted the figures and modified the presentation modes of these figures to improve the quality of all figures in the revised manuscript.

Question 2: In introduction part, the contributions of the paper should be rewritten clearly.

Response 2: Thank you for this valuable suggestion. We have rewritten the introduction (highlighted in yellow) to explain this manuscript's contributions clearly.

Question 3: The computational complexity of deep-learning-based autoencoder for anomaly behavior detection should be analyzed.

Response 3: Thank you for this valuable suggestion. The time complexity of the proposed autoencoder model was analyzed and evaluated in this revised manuscript. We added the analysis description of time complexity in Section 3.2 and provided the evaluation result of the proposed autoencoder model in Section 4.

Question 4: The proposed deep-learning-based autoencoder should be further compared with more advanced deep learning approaches.

Response 4: Thank you for this important suggestion. Indeed, if this manuscript can use the other more advanced deep learning models to evaluate and compare the performance, it will obtain the complete and comprehensive evaluation results. However, this manuscript aims to prove that the anomaly detection model using the high-level features can get better performance than using the low-level features, and we need more time to establish the other more advanced deep learning models and perform the related experiments. Therefore, we take the issue for future work and point it out in the conclusion section.

Question 5: The author is invited to perform a thorough proofread of their manuscript, as I can still spot some spelling/grammar mistakes in the paper.

Response 5: We checked the revised manuscript carefully to correct the spelling and grammar mistakes. Thank you for pointing out this issue.

Question 6: The literature review about anomaly data detection are not sufficient in the current version of this paper. For example, the following manuscript titled as “Robustness of Short-term Wind Power Forecasting against False Data Injection Attacks” is suggested to be included.

Response 6: Thank you for this valuable suggestion. We have reviewed the recent studies in this research field again and then cited the related papers about the anomaly detection of power consumption in the second section to provide a complete background review. Additionally, the paper you suggested was also cited in this revised manuscript.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Accept in present form

Reviewer 2 Report

no comments