An Intelligent Location Method for Power System Oscillation Sources Based on a Digital Twin
Round 1
Reviewer 1 Report
Since more and more renewable energy sources are connected to the power grid the probability of system oscillations caused by power electronic devices increases significantly. That is why it is necessary to localize the oscillation source. For this purpose, in this paper a digital twin-based solution for power system oscillation source localization is proposed, for problems occurring in new power systems.
An intelligent localization algorithm for the oscillation source is constructed by combining the branch potential function and the data-driven fusion.
The authors propose the use of very recent techniques, based on studies published in the last two years. The references underline this aspect.
However, the authors are asked to provide more explanations regarding the techniques used, especially in the theoretical description in the first part (sections 2 and 3). Also, the authors should clearly specify their own contributions, delimiting them better in relation to the elements taken from the recent bibliography.
Section 1 describes quite generally the information in the field, without too many details and without very clearly emphasizing the new elements brought by the work.
Section 2 is extremely short; it offers a superficial presentation, mainly taken from the bibliography, without giving sufficient explanations.
Section 3 is more detailed, but it is not clear what is original contribution and what is taken from the bibliography.
Abbreviations are used, not all explained.
The paper addresses an important and current topic, but for a better understanding of the paper it is necessary to add some additional explanations.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
In this paper, the authors propose a new method to determine the source of oscillations in power systems using their digital twin. The relevance of the considered problem is undoubted, since, on the one hand, in modern power systems with an increasing amount of RES generation there is an urgent issue of occurrence of oscillations in a wide range of frequencies, and, on the other hand, there is an active development of approaches to the implementation of power system digital twins. The paper is well-structured. The paper is of interest of the journal, but there are some comments to authors before the paper could be accepted.
1. The authors’ contribution is not clear enough. Please highlight the contribution of this paper and appropriate particular study as a list (like in these refs: (1) https://doi.org/10.1109/TPWRS.2021.3059197; (2) https://doi.org/10.1109/TPWRS.2020.3001919).
2. The Introduction section is too short. The authors consider two problems – a method for determining the source of oscillations and the use of a digital twin of the power system for this purpose. The Introduction gives an overview of the first task only. It is recommended to supplement this section also with an analysis of existing approaches to the implementation and application of digital twins of power systems (e.g., https://doi.org/10.1109/ACCESS.2022.3217228; https://doi.org/10.1109/TPWRS.2021.3063511; https://doi.org/10.1109/TII.2023.3244329, etc.).
3. The approach proposed by the authors is based on the use of CloudPSS XStudio platform. However, this paper does not provide information about the operation of this simulator: (1) What is the computational core? (Fig. 5 shows that the graphical representation of the model is similar to PSCAD); (2) What models are used in the calculation? (the level of detail of the models determines the quality of process simulation)
4. Besides, this paper does not provide information on the scope of application of the proposed approach – for what range of oscillation frequencies it can be used, how does it change the level of accuracy?
5. In this paper, the authors also showed only the accuracy of oscillation source location, but did not provide information on how long it takes to identify the source. This information is very important for the system operator to take the necessary measures to eliminate oscillations in the power system.
6. In Section 5, the authors present the research results of the proposed approach. At the same time, the simplest power system scheme is considered. In general, in such a power system the source of oscillations can be determined by more simple methods. Thus, the considered model can be used only for generalized validation of the performance of the proposed approach. Perhaps it is necessary to consider the operation of the approach in a more complex scheme with more nodes and sources (e.g., at least IEEE-39)?
7. In Table 2, the authors indicate the source of oscillations at the location of PMU0. However, it is not clear either from the diagram (Fig. 9) or from the text where PMU0 is located?
8. The text of the paper should be carefully proofread and grammar and orthographic errors should be corrected (e.g., in description of Fig. 9).
The text of the paper should be carefully proofread and grammar and orthographic errors should be corrected (e.g., in description of Fig. 9).
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
The authors propose a power system oscillation source localization based on a digital twin-based solution. The paper introduction is well written indicating the current techniques of oscillation localization. However, the paper title is a little confusing, mixing the proposed localization technique and also the system that was developed with this propose. I suggest the authors review the paper title.
The section on Acquired Signal Preprocessing, introducing the Generative Adversarial Imputation Network (GAIN) used to retrieve the data is well written. However Figure 2 should be better explained in the text, trying to highlight each step of Figure 2 with the text. Basically it seems that Figure 2 is inserted in the paper without any context/link with the text.
Is the rotor angle derivative equation (Eq6) correct? Shouldn't the derivative of the rotor angle just be the rotor speed?
Equation 8 should be better explained in the text. Several terms and variables are not clear to the reader.
Is Figure 3 correct? The inputs ("Output of Method 1") appear to be the same. I was unable to correlate Figure 3 with the explanation of the paper.
In Figures 5 and 6, the authors present, from my understanding, the software that was developed to detect the power system oscillation sources. The development of the software is a little confusing. I suggest rewriting this part trying to clarify what the detection technique itself is and what the software is.
Finally, here are my comments on the results presented:
The authors show that the proposed GAIN technique has a greater ability to reconstruct lost data. However, the test was performed only for one dataset. I suggest analyzing more scenarios, with different data and different degrees of information loss (and not just 30%). Thus, the validation of the GAIN technique will be validated in multiple scenarios.
The authors present, separately and jointly, the techniques of "Branch Potential Energy Function" and "Data-driven Combined with Branch Potential Energy". Indicating that there is a substantial improvement in the localization of the oscillating source when the two techniques are used together. Again, this conclusion was validated in only one scenario. I suggest that the system presented in Figure 9 be modified (different generation and load levels, for example) in order to validate this information.
In addition, it was not clear how the authors arrived at some conclusions and it was also not clear how the system was simulated.
For example, what happens in approximately in 3sec on the system shown in Figure 9 to change the Branch Power Energy of Generators 1-4 (Fig, 10)? How do generators 3-4 have positive power? Does this indicate active power consumption?
The authors make associations of the potential energy of several the branches (lines 381-385) that cannot be associated with Figure 10. Please better explain this analysis by associating the branches with the generators.
Finally the authors state: "The results of the two methods are consistent. It determines that the oscillation source is at PMU0." But how can this be asserted? What is the methodology to reach this conclusion?
In other words, the methodology presented is not sufficient to reach the conclusions that the authors state in the article. Please improve the explanation detailing the results.
I suggest that authors do a thorough review of the paper writing/formatting. Below are some points of the text that need to be improved:
1. There are acronyms in the text that were not explained, such as UI, LSTM, KL.
2. There are typos (lack of spacing), especially when the authors use references, as for example in "method[6]" (line 41), "Super-Resolution[17]" (line 140) and others.
3. The formatting of the variables that are used in the equations are not exactly the same as the formatting of the same variables that are cited in the text. For example the variable "n" (line 303) does not have the same formatting (italic) as the variable "n" in equation (12). Formatting carelessness of this type happens successively in the text.
4. I suggest avoiding the use of personal pronouns in scientific texts, such as "we can" (line 192), "you can" (line 340).
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
The reviewer is grateful to the authors for the work done and for taking into account all the comments in the new version of the paper.
Additional proofreading is nessecary to correct some mistakes (e.g. Nvidia not Mvidia)
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
All my suggestions were addressed. I only ask that the authors correct the branch "2-12-11" (line 437) to "3-12-11".
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
