AI-based solar photovoltaic power forecasting

With the aggravation of the global environmental crisis, renewable energy has achieved unprecedented development. Photovoltaic power generation is one of the most widely used new energy sources because of its clean, safe, pollution-free and other advantages. However, photovoltaic power generation is...

Full description

Saved in:
Bibliographic Details
Main Author: Wang, Yichen
Other Authors: Xu Yan
Format: Thesis-Master by Coursework
Language:English
Published: Nanyang Technological University 2022
Subjects:
Online Access:https://hdl.handle.net/10356/163749
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
Description
Summary:With the aggravation of the global environmental crisis, renewable energy has achieved unprecedented development. Photovoltaic power generation is one of the most widely used new energy sources because of its clean, safe, pollution-free and other advantages. However, photovoltaic power generation is greatly affected by weather factors and has greater randomness, which poses a threat to the safe and stable operation of the power system. Therefore, the dissertation focuses on the research and discussion of photovoltaic power prediction. Firstly, the classification of photovoltaic power generation forecasting is introduced in terms of technologies and forms, and the literature review of photovoltaic power generation forecasting based on artificial intelligence is carried out in terms of machine learning and deep learning. The existing research methods such as MLP, SVM, LSTM, CNN, and so on are listed. Secondly, components of photovoltaic prediction model are introduced, the principles of data preprocessing algorithm, artificial intelligence prediction algorithm and formulas of evaluation indicators used in this dissertation are analyzed. VMD, Pearson correlation coefficient, DenseNet, LSTM, XGBoost and so on have been applied in this dissertation. Finally, prediction results and errors of the three used data sets are listed and discussed. Conclusion that which method is the most suitable for each dataset are drawn as well. Key Words: artificial intelligence; solar power forecasting, DenseNet, LSTM, SVM