A Novel Hybrid Predictive Model for Ultra-Short-Term Wind Speed Prediction

Round 1

Reviewer 1 Report

The paper deals with the ultra-short-term forecasting of wind speed through a novel more accurate hybrid model. The proposed approach combines improved complete ensemble empirical mode decomposition with adaptive noise, the sample entropy, optimized recurrent broad learning system, and broadened temporal convolutional network. The obtained results seem reliable and accurate. The performance of the proposed model was compared with twelve advanced predictive models; it shows a better forecasting accuracy in all four study cases.

 

The following issues are recommended to improve the paper:

1.     “…wind energy provides unstable power due to its intermittent and fluctuating nature. …. Accurate short-term wind speed prediction (WSP) can largely avoid this problem.” Please justify with scientific arguments this statement!

2.     “Wind speed forecasts can be classified into ultra-short-term forecasts, short-term forecasts, mid-term forecasts, and long-term forecasts”. For the sake of clarity, quantitative information is needed to be associated to this qualitative classification, or refer here relevant work(s).

3.     Define all symbols at their first use, even they are well known in literature (or introduce them in Nomenclature). E.g., μ, σ, etc.

4.     The Sections 2.2, 2.3, and 2.4 should be developed in a more detailed way by clearly defining the symbols and rationale of the relations. E.g., explain why the maximum value of  ?  is 11 (line 303).

5.     Table 1. Check the measurement units, it’s a realistic wind speed of 560 m/s (i.e., ~ 2000 km/h)? Define the altitude of these speed measurements.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

This paper proposes a hybrid methodology for ultra-short-term wind speed prediction (15 minutes prediction time horizon). Specifically, the proposed methodology involves time series decomposition approaches and a deep learning method as final prediction steep: The improved complete ensemble empirical mode decomposition with adaptive noise (ICEEMDAN), the sample entropy (SE), optimized recurrent broad learning system (ORBLS), and broadened temporal convolutional network (BTCN) form the final hybrid algorithm. 

I have seen this type of approaches before in the literature, in fact the authors cite some similar approaches, some of them even recently published in Energies:

Liu, J.; Shi, Q.; Han, R.; Yang, J. A Hybrid GA–PSO–CNN Model for Ultra-Short-Term Wind Power Forecasting. 506
Energies 2021, 14, doi:10.3390/en14206500.

Wang, J.; Yang, Z. Ultra-short-term wind speed forecasting using an optimized artificial intelligence algorithm. 476
Renewable Energy 2021, 171, 1418-1435, doi:10.1016/j.renene.2021.03.020.

Nie, Y.; Liang, N.; Wang, J. Ultra-short-term wind-speed bi-forecasting system via artificial intelligence and a dou- 495
ble-forecasting scheme. Applied Energy 2021, 301, doi:10.1016/j.apenergy.2021.117452.

Zhang, Y.; Han, J.; Pan, G.; Xu, Y.; Wang, F. A multi-stage predicting methodology based on data decomposition 513
and error correction for ultra-short-term wind energy prediction. Journal of Cleaner Production 2021, 292, 514
doi:10.1016/j.jclepro.2021.125981.

There are more, some of them very similar to that proposed in this work:

Shang, Y., Miao, L., Shan, Y., Gnyawali, K. R., Zhang, J., & Kattel, G. (2022). A Hybrid Ultra-short-term and Short-term Wind Speed Forecasting Method based on CEEMDAN and GA-BPNN. Weather and Forecasting, 37(4), 415-428.

Jiang, Z., Che, J., & Wang, L. (2021). Ultra-short-term wind speed forecasting based on EMD-VAR model and spatial correlation. Energy Conversion and Management, 250, 114919.

Tian, Z., & Chen, H. (2021). A novel decomposition-ensemble prediction model for ultra-short-term wind speed. Energy Conversion and Management, 248, 114775.

Sibtain, M., Bashir, H., Nawaz, M., Hameed, S., Azam, M. I., Li, X., ... & Saleem, S. (2022). A multivariate ultra-short-term wind speed forecasting model by employing multistage signal decomposition approaches and a deep learning network. Energy Conversion and Management, 263, 115703.

In fact, there are so many articles in this topic, all them proposing "novel" approaches that it is not possible to check out what are really the best approaches. All them are compared with alternative approaches, but I haven't seen a real comparison of these approaches among them, since all them claim to be the best of the best. First, all these approaches are only based on the wind speed series, without external variables to characterize the atmosphere state. It could make sense, since prediction to only 15 minutes time horizon is extremely short-time, and it is supposed that the atmosphere state is unvariable. However, I do think that the inclusion of variables measuring what is happening in the boundary layer would help improve the prediction in all these papers above. Very local variables such as the vertical wind velocity, or other variables able to measure the turbulent energy transport should improve the performance of these systems.

Regarding this specific paper, I wouldn't support the acceptance of "just another algorithm" in this topic, unless the authors are able to show that their proposal is worth to be published, by means of a serious comparison with some of the approaches proposed before (some of the papers reference above), and show advantages. In its current form, the experimental part of the paper does not show a real improvement over some hybrid approaches previously published. The other option is that the authors incorporate some boundary layer variables and check out the improvement of their approach in ultra-short time prediction with these new variables. This would be a real innovation and novelty for the paper. Any other thing than a serious review of the paper is nothing but an unfortunate waste of time.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

The authors have answered well to my concerns, including significant comparison with recent articles, where they show their proposal is competitive. I only have a comment regarding the revised submission, about the new author included: The authors have not commented in their response what is the reason why they have added a new authors (or at least I haven't seen any comment about this point). The authors should explain the reason of including a new author at this step, and what has been the contribution of the new author. Also, the authors must explain why they have highlighted in red a previous existing author, is it because they have changed his affiliation? Please explain.