摘要(英) |
Solar power is gradually becoming the main force of renewable energy in Taiwan. As the proportion of solar power generation increases, the challenging task of maintaining the stability of the grid becomes crucial. Therefore, for the development of smart grids, solar power generation forecasts has become an important issue.
The dataset in this thesis comes from the power generation data of multiple inverters monitored in a solar power system. It is desirable to design a multivariable time series to forecast the generated power of multiple inverters. Traditional modeling methods lead to relatively high computational costs and difficulties in capturing multivariate dependencies of data. Therefore, we propose a deep learning model, which combines convolutional neural network and long- short term memory recurrent network to predict the day-ahead hourly power generation. In this model, the convolutional neural network is used to extract the features of multivariate time series, and the long-short term memory network predicts the multivariate results.
In order to make the results more accurate, we use the data of Numerical Weather Prediction and perform feature selection to combine the weather features to train the model. The proposed method further separates the sunny and rainy models according to the rainfall condition of forecast day. Experiments show that the combination of weather information can further reduce the errors. Finally, considering the difference in weather conditions between the northern and southern parts of Taiwan, experiments involving multiple sites in different regions were evaluated to verify that the combination of weather information can be widely used in regions with different weather conditions in Taiwan. |
參考文獻 |
[1]2021太陽能趨勢|太陽能還有發展性嗎?解析台灣、國際綠能之路。取自https://blog.pgesolar.com.tw/2021/02/08/%E5%A4%AA%E9%99%BD%E8%83%BD%E8%B6%A8%E5%8B%A2
[2]智慧電網介紹。取自https://smartgrid.taipower.com.tw/en/index.aspx
[3]Pedro, Hugo TC, and Carlos FM Coimbra. "Assessment of forecasting techniques for solar power production with no exogenous inputs." Solar Energy 86.7 (2012): 2017-2028.
[4]Fentis, Ayoub, et al. "A Machine Learning Based Approach for Next-Day Photovoltaic Power Forecasting." 2020 Fourth International Conference On Intelligent Computing in Data Sciences (ICDS). IEEE, 2020.
[5]Suresh, Vishnu, et al. "Forecasting solar PV output using convolutional neural networks with a sliding window algorithm." Energies 13.3 (2020): 723.
[6]Sheng, Hanmin, et al. "Short-term solar power forecasting based on weighted Gaussian process regression." IEEE Transactions on Industrial Electronics 65.1 (2017): 300-308.
[7]Ordiano, Jorge Ángel González, et al. "Photovoltaic power forecasting using simple data-driven models without weather data." Computer Science-Research and Development 32.1-2 (2017): 237-246.
[8]Alomari, Mohammad H., Jehad Adeeb, and Ola Younis. "Solar photovoltaic power forecasting in jordan using artificial neural networks." International Journal of Electrical and Computer Engineering (IJECE) 8.1 (2018): 497-497.
[9]Zhu, Honglu, et al. "An improved forecasting method for photovoltaic power based on adaptive BP neural network with a scrolling time window." Energies 10.10 (2017): 1542.
[10]Aprillia, Happy, Hong-Tzer Yang, and Chao-Ming Huang. "Short-Term Photovoltaic Power Forecasting Using a Convolutional Neural Network–Salp Swarm Algorithm." Energies 13.8 (2020): 1879.
[11]Tang, Xianlun, et al. "Short-term power load forecasting based on multi-layer bidirectional recurrent neural network." IET Generation, Transmission & Distribution 13.17 (2019): 3847-3854.
[12]Suresh, Vishnu, et al. "Forecasting solar PV output using convolutional neural networks with a sliding window algorithm." Energies 13.3 (2020): 723.
[13]Li, Gangqiang, et al. "Recurrent neural networks based photovoltaic power forecasting approach." Energies 12.13 (2019): 2538..
[14]Lai, Guokun, et al. "Modeling long-and short-term temporal patterns with deep neural networks." The 41st International ACM SIGIR Conference on Research & Development in Information Retrieval. 2018.
[15]Cheng, Jiezhu, Kaizhu Huang, and Zibin Zheng. "Towards better forecasting by fusing near and distant future visions." Proceedings of the AAAI Conference on Artificial Intelligence. Vol. 34. No. 04. 2020.
[16]Zhen, Zhao, et al. "Deep learning based surface irradiance mapping model for solar PV power forecasting using sky image." IEEE Transactions on Industry Applications 56.4 (2020): 3385-3396.
[17]Shi, Jie, et al. "Forecasting power output of photovoltaic systems based on weather classification and support vector machines." IEEE Transactions on Industry Applications 48.3 (2012): 1064-1069
[18]Acharya, Shree Krishna, Young-Min Wi, and Jaehee Lee. "Day-Ahead Forecasting for Small-Scale Photovoltaic Power Based on Similar Day Detection with Selective Weather Variables." Electronics 9.7 (2020): 1117.
[19]Cortes, Corinna, and Vladimir Vapnik. "Support-vector networks." Machine learning 20.3 (1995): 273-297.
[20]Ahn, Hyung Keun, and Neungsoo Park. "Deep RNN-Based Photovoltaic Power Short-Term Forecast Using Power IoT Sensors." Energies 14.2 (2021): 436.
[21]K Fukushima. “Neocognitron: A self-organizing neural network model for a mechanism of pattern recognition unaffected by shift in position.” Biological cybernetics, 36(4), 193-202, 1980.
[22]LeCun, Yann, et al. "Gradient-based learning applied to document recognition." Proceedings of the IEEE 86.11 (1998): 2278-2324.
[23]池化層示意圖、圖片來源取自https://www.quora.com/What-is-the-benefit-of-using-average-pooling-rather-than-max-pooling
[24] Chung, Junyoung, et al. "Empirical evaluation of gated recurrent neural networks on sequence modeling." arXiv preprint arXiv:1412.3555 (2014).
[25]長短期記憶模型示意圖、圖片來源取自colah’s blog《Understanding LSTM Networks》:http://colah.github.io/posts/2015-08-Understanding-LSTMs
[26]太陽能電網示意圖,圖片來源取自http://gec.jp/jcm/projects/18pro_ken_01
[27]Bias–variance tradeoff,圖片來源取自https://en.wikipedia.org/wiki/Bias%E2%80%93variance_tradeoff |