A Comprehensive Application of Machine Learning Techniques for Short-Term Solar Radiation Prediction

Round 1

Reviewer 1 Report

This paper is focused on an interesting topic, related with renewable energy, particularly solar panels, photovoltaic, etc. It is written in proper English and pleasant to read. The introduction part is very informative, with a good mathematical background. Whereas, the research part describes a broad campaign. The number and amount of gathered and processed data is sufficient. However, I do encourage the Authors to present their results in other forms than just plain tables. The number and quality of cited references is sufficient.

Overall, this paper requires modifications before it can be accepted and published. It is a good paper, however it deserves to be a very good one. In order to raise its quality, acquaint with provided suggestions and comments.

Suggestions and comments:

• Fig. 1. – check the spacing between words [Hybrid methods].
• Minor editorial and/or formatting issues, e.g. additional (unnecessary) or lack of space sign (on multiple pages), different fonts and/or spacing between following lines (see line 108-109),
• The word [daily] is highlighted multiple times (see e.g. page 4) – why is it?
• Figure captions should be inserted just after the figure itself and on the same page (see the situation with Figs. 3, 10, and make necessary corrections).
• Formatting of mathematical equations and formulas – they are not uniformly prepared (see e.g. page 6-8). Use the build-in editor, use the same setting, fonts, etc.
• Check the formatting of Chapter 5 and sub-chapter 5.1. (lines 401-402) and correct them.
• Authors mention multiple times in text that there were [pieces of missing data / missing values]. What was the cause of it? What is the effect of it? Why are those pieces (values) of missing data variable (unequal to each other)? An appropriate commentary in text should appear.
• Furthermore, please specify what was the source of your data. How was it obtained (sourced database or own gathered database), and in what way (from a government or non-government institution, own survey, what devices/sensors/nodes were utilized). Additionally, provide at least basic info concerning utilized equipment.
• Figs 10 and 11 present plots with multiple color. What does each color stand for? Additional info should be provided both in text and on graph.
• Check the alignment of Tables 1-12, notice paragraph indentation and make necessary corrections.
• Consider extending the Conclusions section. Surely Authors have more thoughts and observations based on carried out work that may be valuable for numerous potential readers.

Author Response

Please see the attachment.

Reviewer 2 Report

The authors study the forecasting of the output power of photovoltaic systems, which is required for the proper operation of the power grid and optimal management of the energy flows occurring in the photovoltaic system. It is inferred from the solar radiation data in Jiangsu, China over a period of two years. A way to recover missing data by matrix padding is under consideration. To reduce the effect of the weather on solar radiation, a k-nearest-neighbor method is used to divide the data into three clusters. For each cluster, four artificial neural networks are trained to predict short-term solar radiation. However, there are some minor inaccuracies in the article:
1. Figure 1: only 4 machine learning methods are listed;
2. Figure 3: Description is on the next page;
3. In section 4.1 BP neural networks, citations are required.

Author Response

Please see the attachment.

Author Response File: Author Response.doc