Predictive Analysis and Correction Control of CCT for a Power System Based on a Broad Learning System
Round 1
Reviewer 1 Report
The transient stability of power systems is crucial for their stable operation. However, as the power grid scale expands and the voltage level increases, transient stability problems become more pronounced. The delay in the operation of relay protection devices, especially when their operation time exceeds the critical clearing time (CCT), poses a risk to system stability. As such, it's vital to accurately predict and adjust CCT. This paper presents a method to predict, evaluate, and adjust CCT using the Bi-Level Selection (BLS) model. The BLS model allows easy calculation of the CCT sensitivity and correction of CCT to enhance system stability. The proposed method was tested on a 4-machine 11-bus system and a 10-machine 39-bus system, with the results indicating BLS's effective performance in predicting, evaluating, and correcting CCT. My comments on this work are
1. The authors have tested their approach on a 4-machine 11-bus system and a 10-machine 39-bus system. While these systems provide some level of validation, they may not represent the full complexity or diversity of real-world power systems.
2. The paper refers to the limitations of various other AI models, such as Artificial Neural Networks (ANN) and Lasso, but does it provide a rigorous comparative analysis to illustrate how BLS outperforms these models in predicting and correcting CCT?
3. The paper mentions that the BLS model can adjust the CCT when it falls below the operation time of the relay protection device. However, does the paper discuss how the model can adapt to changes in the grid and how it can handle unforeseen disturbances?
4. As with most machine learning models, the BLS algorithm heavily depends on the tuning of hyperparameters. While the paper does mention that BLS is designed to overcome issues related to the hyperparameters of deep learning models, it doesn't seem to address how the hyperparameters of the BLS model itself are tuned, and how they affect its performance.
5. The paper does not describe the error metrics used to evaluate the performance of the BLS. It's important to understand how the accuracy of the model is determined and how it handles errors.
6. The quality of all figures is very very poor, the authors should first read the figures, without improving the quality of figures this paper has no worth.
7. The paper sets the relay protection action time, which serves as the threshold value, to be 0.2 seconds. However, it doesn't provide any justification for why this particular value is used. In real-world applications, the appropriate value for this threshold could vary based on a number of factors.
8. The paper discusses the sensitivity of CCT to a single variable (i.e., active power of the generator). This could potentially oversimplify the problem, as the stability of power systems is typically influenced by multiple variables, not just one. The model might not account for interactions and dependencies between different variables.
9. The paper assumes that only the active power of the generator can be changed to improve CCT. This might be a limiting assumption, as there could be other control variables that can be manipulated to improve CCT.
10. The paper does not mention any comparison with other existing methods. A comparative analysis could strengthen the argument for the proposed models' effectiveness.
11. The random variation ranges for each node parameter are specified as the reference value multiplied by a factor within 0.8–1.2. It's unclear how these ranges were determined, and they may not accurately reflect real-world variability. This might limit the model's applicability to broader situations.
12. The random variation ranges for each node parameter are specified as the reference value multiplied by a factor within 0.8–1.2. It's unclear how these ranges were determined, and they may not accurately reflect real-world variability. This might limit the model's applicability to broader situations.
13. Although the paper provides several tables, figures, and results, it doesn't adequately discuss or interpret these results. There's a need for more in-depth discussion and analysis to fully understand the implications of the findings.
14. The text states that the BLS model can quickly update system weights without complete retraining when new data is added. However, it does not provide any information about the computational efficiency of these processes. In a real-world application, where timely response is critical, computational efficiency can be a significant concern.
1. Some sentences are quite long and complex, which can make them difficult to read and understand. Breaking these up into shorter sentences might improve readability. 2. There are some formatting errors and missing references, such as "Table 11Error! Reference source not found." These need to be corrected for a polished and professional presentation. 3. There seems to be some inconsistency in the use of certain terms. For instance, "4-machine 11-node system" is also referred to as "4-machine 11-bus system". Consistency in terminology can avoid potential confusion for readers. 4. The use of more precise language could further improve the quality of the text. For example, instead of saying "the BLS model is fast in predicting CCT", specifying how fast (e.g., in terms of processing time) would provide more concrete information. 5. Some information seems to be repeated, such as the explanation of what the BLS model does. Removing these repetitions could make the text more concise. Overall, with some editing for clarity, conciseness, and formatting, the quality of the English could be further improved.
Author Response
We are extremely grateful to reviewer‘s suggestion. Based on your comments, we have revised manuscript. More details are in document.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
This article is pertinent and contributes seriously to the science behind the transient stability of power systems and quick response to system failures using BLS models.
The language quality is clear and sound (very few typos that a quick spell-check could resolve). But I have identified a few items to adjust (see the other Comments on the Quality of English for details) and further target the studies’ methodology and translate them into material clarification and potential recommendations for future research as outlined hereafter:
Line 88: Could the author better outline the differences between their approach and the 2 conventional models, i.e., “time-domain simulation and direct method” concepts, early in the text, such as in the “2.1. Traditional calculation methods of CCT” section when introducing the “Basic Models”? Similar to what is explained in section 2.2.1 regarding the Broad Learning System that minimizes the risks of falling into a DPN “Hyperparameters” bottleneck. This differentiation from the basic models is also key to showing the additional input from the Research and to the readers to understand better the extent of the work presented in the paper. It could also be re-emphasized in the conclusions.
Line 95. In equation (2), some terms are not defined, such as Yij, Bij, or Gij, as soon as they appear in the equations. Please revise and add missing term definitions in this equation and equation 6.
Line 141. In equation (6), ? is not defined, such as for the proceeding item; please revise and add the missing term definitions in this equation.
Line 205: Please clarify what you mean by “This will get the input parameters.” Are the input parameters randomly generated indistinctly, or are they part of the imposed “inputs”?
Line 297: This is an important observation “we propose an optimal adjustment model based on BLS, which adjusts CCT to be higher than or equal to the operation time of the relay protection device at the minimum change of generators”, but it is also a recommendation from the authors’ work. It should be outlined in the conclusions (Line 398 and beyond), justifying that the relay protection device cuts out the faulty line with significant reaction time, affecting the stability of power systems. This might prove helpful to the readers.
In the Conclusions section (Lines 393+), you should outline one of the notable statements, i.e., the main differences between the conventional models and/or the optimal adjustment model based on BLS. It could also outline future research on that topic since more studies might be needed here, and the readers would welcome some orientation or recommendations from the authors.
The language quality is clear and sound (very few typos that a quick spell-check could resolve). But I have identified a few items:
Line 8-11: Please avoid very long sentences (especially in the abstract) with multiple repetitions of the same word in one single sentence, such as in: “The transient stability of power systems is an important factor for the stable operation of 9 power systems, but with the increase in voltage level and the expansion of the power grid scale, the 10 current transient stability problem is particularly prominent.”
Line 46: Please define “critical clearing time” (CCT) on its first appearance in the text and not only in the Abstract.
Author Response
We are extremely grateful to reviewer‘s suggestion. Based on your comments, we have revised manuscript. More details are in document.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
The authors have addressed all the comments.
The english has been imoroved.
