A Multi-Step-Ahead Photovoltaic Power Forecasting Approach Using One-Dimensional Convolutional Neural Networks and Transformer
Round 1
Reviewer 1 Report
Comments and Suggestions for Authors§ What was the extent of data from the DKASC-ASA-1A database? In section 2.1 you mentioned that the last two years were used for testing. So, you could specify the exact extent of the database used.
§ In Table 1, how come you have negative values for Wind speed and Global horizontal radiation minimum values?
§ Figures should be inserted within the manuscript after they are first mentioned (check Figures 1, 2, and 3).
§ I don’t think that functions like “sqrt” and “sum” must be explained, like you did in line 245.
§ You should provide details about the results obtained with your model, and first compare it with the actual results from the DKASC database. Your estimated power, against the real one, for the considered time interval. After that, you should consider comparisons with other models.
§ Also, I think when you compared your results with the ones from other studies, you did it in a misleading manner. For example, from reference [33] you chose to present only the results according to CNN-GRU technique, which are inferior to the results obtained by your model. However, you didn’t mention that the authors of the study cited as [33] proposed their own model, named ESNCNN, which provides better results than yours (RMSE – 0.161 and MAE – 0.0845).
§ Also, in references [32] and [34], the authors used another database for their study, namely DKASC-ASA-1B, where the PV array rating is 23.4 kW. Your study was conducted on the DKASC-ASA-1A, where the generating capacity is 10.5 kW. Thus, you compared your study’s results with results from different studies, with different inputs and different real results. Hence, the validity of your study results is questionable.
Author Response
We would like to extend our sincere thanks to Reviewer 1 for your insightful and detailed review. Your comments were invaluable in improving our work, and we sincerely appreciate the time and effort you took to evaluate our submission. Please find the corrected file and our response attached, and thank you again for your support.
Author Response File: Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsManuscript misses any details on what is exactly hapening. The announcement of "extensive preprocessing methods and conducted 18
extensive experiments" are not reflected in the manuscript.
What is shown in figure 2 and 3 stays as a "colored box" representation, missing any valuable information on what is happening in the boxes (what is a "Masked Parallelized Attention Mechanism"and which role does it play).
Comparisons with other forecast shemes is limited to bare information on MAE and RMSE, whithout any hints on how the proposed scheme exactly gains its advantages.
Comments on the Quality of English Language
Quality of english mostly ok.
Author Response
We would like to extend our sincere thanks to Reviewer 2 for your insightful and detailed review. Your comments were invaluable in improving our work, and we sincerely appreciate the time and effort you took to evaluate our submission. Please find the corrected file and our response attached, and thank you again for your support.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsCongratulations on your work!