A Hybrid Generative Adversarial Network Model for Ultra Short-Term Wind Speed Prediction
, Longnv Huang 1,†, Jiehui Huang 1,†, Qiaoan Liu 2, Limin Chen 1,*, Yin Liang 1, Peter X. Liu 1,3 and Chunquan Li 1
Round 1
Reviewer 1 Report
The paper proposes a hybrid generative adversarial network model, which effectively improves the accuracy of ultra-short-term wind speed prediction. Using a hybrid model, combining CEEMDAN, OBLS and WGAN: After the data is decomposed and preprocessed by the CEEMDAN method with reduced wind sequence noise, the proposed HGANN network is input. The network has OBLS as the generator and WGAN's discriminator as its discriminator, and OBLS uses an improved particle swarm algorithm to optimize the hyperparameters of the BLS. Based on the actual data and compared with other common prediction models, the effectiveness of this hybrid generative adversarial network model is verified by four evaluation indexes. The workload of the article is substantial, the simulation is reasonable, and the results are credible.
However, the paper still has the following problems, and it is recommended to revise:
1. The introduction of BLS is not natural and sufficient. It is suggested to highlight the comparison between BLS and neural network to prove the necessity of this improvement work.
2. In the “Generator” section of Figure 1, the annotation in the blue box should be “Enhancement Nodes” instead of “Enhancement Notes”.
3. It is recommended that the title and figure of Figure 3 be placed on one page.
4. It is suggested to add the value of multi-step prediction in wind speed prediction to highlight the necessity of Experiment II.
Author Response
"Please see the attachment."
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Dear authors,
The topic of your manuscript “A Hybrid Generative Adversarial Network Model for Ultra Short-term Wind Speed Prediction” is interesting. I have the following comments,
The manuscript has the structure and narrative of a technical report more than a research paper. The structure and narrative should be improved following the traditional sections of a research paper: introduction, methods & materials, results, discussion and conclusion.
The manuscript has high content on the introduction and methods sections but the results and discussion sections are limited. I would suggest expanding the results and introducing a discussion section.
The study considers data from Germany for one month (Aug 23 2019 to Sep 22 2019). Why was this period considered adequate ? Wind speed is highly variable by geographical locations and diverse time periods. Is this one-month data representative for other time periods or other locations? Is it possible that considering other time periods or other locations (with higher variability) would generate different results? Is the proposed model only useful for Germany and for the particular wind speed variability behavior on that particular time frame or can it be expanded to more general settings? How can this be validated and justified? This needs to be justified and discussed to validate the use of the indicated data set.
Bullet points and numbered points are frequently used in the manuscript in different sections. This is more adequate for a technical report. A narrative style with a proper storytelling would be more adequate for the manuscript, substituting all bullet points and numbered statements.
Figure 1 quality should be improved. It is very difficult to read. It should be described with more detail, connecting with the narrative of the manuscript.
It seems that results presented in Figure 3 and Table 4 are the same. What is the difference between them and why are both necessary?
Figure 4 is too crowded and difficult to read. Quality should be improved.
Descriptions for the methods on Section 2.2, 2.5 and Algorithm 1: listing steps and algorithm functionality are more adequate for a technical report. Converting this information to flow charts with strong description on their functionality would improve the manuscript.
The use of numbered statements on the Conclusion makes the manuscript appear like a technical report. A narrative style with a proper storytelling would be more adequate for the manuscript, substituting all bullet points and numbered statements.
I would strongly suggest for the manuscript to be reviewed and edited by a proficient native English speaking editor after all the previous comments have been addressed.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
To improve the accuracy of ultra-short-term wind speed prediction, a hybrid generative adversarial network model (HGANN) is proposed in this paper. Firstly, to reduce the noise of the wind sequence, the raw wind data is decomposed using complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN). Then the decomposed modalities are entered into the HGANN network for prediction. HGANN is a continuous game between the generator and the discriminator, which in turn allows the generator to learn the distribution of the wind data and make predictions about it. Notably, the authors developed the optimized broad learning system (OBLS) as a generator for the HGANN network, which can improve the generalization ability and error convergence of HGANN. Furthermore, improved particle swarm optimization (IPSO) was used to optimize the hyperparameters of OBLS.
The paper idea is valid and adds knowledge to the literature. The paper is related to the journal scope. Overall, the paper needs major corrections.
The limitation of the paper lies in paper structures and its presentation as it has to be improved.
· The first section (Introduction) is too long. It should be divided into 2 sections (1. Introduction) and (2. Background).
· The related works are missing and a new recently published paper has to be cited, e.g:
o Palanisamy, S.; Thangaraju, B.; Khalaf, O.I.; Alotaibi, Y.; Alghamdi, S. Design and Synthesis of Multi-Mode Bandpass Filter for Wireless Applications. Electronics 2021, 10, 2853.
o Khodayar, M., Wang, J., & Manthouri, M. (2018). Interval deep generative neural network for wind speed forecasting. IEEE Transactions on Smart Grid, 10(4), 3974-3989.
o Wang, F., Tong, S., Sun, Y., Xie, Y., Zhen, Z., Li, G., ... & Liu, D. (2022). Wind process pattern forecasting based ultra-short-term wind speed hybrid prediction. Energy, 124509.
o Zhen, Z., Qiu, G., Mei, S., Wang, F., Zhang, X., Yin, R., ... & Catalão, J. P. (2022). An ultra-short-term wind speed forecasting model based on time scale recognition and dynamic adaptive modeling. International Journal of Electrical Power & Energy Systems, 135, 107502.
o Anuradha D, Subramani N, Khalaf OI, Alotaibi Y, Alghamdi S, Rajagopal M. Chaotic Search-and-Rescue-Optimization-Based Multi-Hop Data Transmission Protocol for Underwater Wireless Sensor Networks. Sensors. 2022; 22(8):2867. https://doi.org/10.3390/s22082867
o Palanisamy, S., Thangaraju, B., Khalaf, O. I., Alotaibi, Y., Alghamdi, S., & Alassery, F. (2021). A Novel Approach of Design and Analysis of a Hexagonal Fractal Antenna Array (HFAA) for Next-Generation Wireless Communication. Energies, 14(19), 6204.
· The discussion is missing and it has to be added before conclusion.
· The paper has to be proofread.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
I think most of review comments were adopted and resolved in the new version of the manuscript. But what I 'm still confused about is why MAPE data in Table 3 and Table 4 is larger than 1 when MAE data is smaller than 1 according to Eq. 18 and Eq. 19. Please give a reasonable explanation.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 2 Report
Most of the review comments were addressed in this new version of the manuscript.
I still suggest for the manuscript to be framed in the traditional sections of a research paper: introduction, methods & materials, results, discussion and conclusion. It would improve narrative style and provide readers with a clearer description of the research being presented.
I suggest expanding the discussion and conclusion which are still limited.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
The authors have considered all my commands and thus the paper can be accepted in current form.
Author Response
Please see the attachment.
Round 3
Reviewer 1 Report
This version is acceptable.
