Impact of an ML-Based Demand Response Mechanism on the Electrical Distribution Network: A Case Study in Terni

Round 1

Reviewer 1 Report

This paper studied an interesting topic and presented an Artificial Intelligence (AI) models for load and generation prediction and grid optimization. After reviewing, some revisions are needed.

1.        There are too many abbreviations in the introduction, making the reading difficult. I suggest a abbreviation list in a separate section.

2.        The authors said that ‘The literature on the topic of DR is quite extensive…’. However, the difference between this article and previous studies is not clearly given.

3.        The reason of ‘modelling and testing an ML-based forecasting model’ is not illustrated in the introduction. Specifically, the ‘ML’ appeared in the contributions and the ‘AI’ appeared in the end of the introduction are confusing.

4.        The authors proposed a ML model and only tested its performance on the basis of a real case. I think this is not enough. It is better to validate the effectiveness of the proposed method by comparing it with other advanced algorithms.

5.        What is ‘MSE’?

6.        The abbreviation ‘ML’ should be given where it first appears.

7.        The parameter ? may have main effect on the optimization results, which is missing in the Results section.

 Minor editing of English language required

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

1.       Avoid using acronyms in the abstract.

2.       What is MG, it should be explained when it first appears.

3.       The Introduction should be summarized to show the research gap before introducing the contributions.

4.       “…simulation of the implementation of an original DR mechanism on a real network”, are you sure that this has not been done in previous studies?

5.       “an ML-based forecasting model” can this be more specific, and it should be mentioned in the Introduction.

6.       “The equations to calculate SSR and SCR are as follows.” References should be added.

7.       What is the novelty of the proposed ML model?

8.       What are the criteria to test the performance of the proposed ML model?

9.       “Additional information about ASM DN can be found in [33] -[35].” This should be more specific, which can help readers to find the sources of data directly and easily.

10.     Quality of Figure 4 should be significantly improved. Same issue for Figures 5 and 6.

11.     More deep analysis should be provided. E.g., In Fig 6, domestic loads are shown before and after being optimized, what are the differences, and the reasons.

12.     Some comparison of the proposed method to current methods should be added, which also shows the novelty and contributions of this paper.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 3 Report

This paper studies the impact of an AI-based demand response mechanism on the electrical distribution network based on a case study in Terni. Some comments and suggestions to further improve quality of the paper are given below:

(i) This paper studies two types of customers including domestic and industrial loads. However, only the load distribution of domestic customers is given in Fig. 3. Why is the load distribution of industrial customers not provided, and what are the different characteristics between domestic and industrial loads. Related figure and discussions need to be added.

(ii) In Section 2.2.2, only the higher loads are retained to simplify the optimization model. However, it should be noted that large loads do not necessarily possess the stronger demand response capability. The demand response capability should be taken as an important criterion in the load selection.

(iii) In the section of “Discretization of states”, the states are discretized on an hourly basis to be applied to the optimization model, so the demand response optimization is carried out with the time interval of 1 hour. Would it help to improve the model accuracy to further reduce the time interval?

(iv) In Section 2.2.3, the power demand load is described with a 15-miniute interval. Why is this time interval shorter than 1 hour, and how can the 15-miniute interval load data be applied to the discretization of states on the 1-hour basis?

(v) With equation (3), the method to tune the α parameter may be described in detail.

(vi) The Results Section is rather insufficient in content and capability to support conclusions. The Figures are not explained in detail and the improved demand response capability is not well presented. Expansion of the result section is needed.

Moderate editing of English language required.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 4 Report

Good day and congratulations on the research. The work was interesting to read.

The paper is quite short and the key concern relates to the research proposition, the paper is quite operational. 

the paper is missing a detailed literature review. This should cover all major themes that contribute towards the research. 

the method is very operational. please provide a theoretical grounding and a knowledge based grounding for the method. the method must elucidate details of theories and literature as grounded. 

the case study and results must be grounded and must be better unpacked.

the research value of this study must be justified and explained

Please review for language

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

All comments have been well replied. I suggest accepting the current version.

Minor editing of English language required

Author Response

Thanks to the reviewer for his/her comments

Reviewer 2 Report

1. “Zakariazadeh et al. [19] introduce” should be “Zakariazadeh et al. [19] introduced”, use the past tense for literature review.

2. “in [20], [21], the correlation between grid type” what are the differences of these two papers?

3. “As described previously, in the literature there is a shortage of articles dealing with DR mechanisms ML-based, applied to a real DN.” What are the challenges, why there is a shortage of this type of article?

4. What are the units for x-axis and y-axis of Figure 8? Same issue for Figure 9.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 3 Report

The comments by the reviewer have been properly addressed. No further comments.

Author Response

Thanks to the reviewer for his/her comments