隨著對於可再生能源的意識逐漸提高，世界各國開始關注碳排放相關之議題。可再生能源是當前全球朝向低碳排放目標的能源解決方案之一。然而，對於發電廠來說，可再生能源的不穩定性會導致發電成本的增加。微電網若有能力準確預測太陽能發電的情形將有助於公司規劃電力供應與儲存，並針對用電需求進行應變，此外，不同的預測範圍與粒度對於發電廠而言，也將在不同層面的規畫提供一定程度的輔助。
然而，天氣和太陽輻射是影響太陽光伏發電最關鍵之因素。天氣因子的多樣性與複雜度高，需要對天氣進行分析，才能找到影響預測的基本因素。本研究使用主成分分析（PCA）、隨機森林和 XGBoost（eXtreme Gradient Boosting）作為特徵工程方法，並分別進行比較。預測之演算法由長短期記憶神經網絡（LSTM）作為延伸，包括構建兩個堆疊的 LSTM 模型：Bi-LSTM（Bi-directional LSTM）和 Auto-LSTM（AutoEncoder LSTM）。本研究分別從兩個來源蒐集天氣數據和太陽能發電數據。數據資料以小時為單位，進行一天後（24小時）與一周後（168小時）的多步預測。
最後，本研究提出一種兩階段之太陽能光伏多步預測模型，以天氣特徵先進行太陽輻射預測，再將太陽輻射預測作為預測太陽光伏預測之資料集，並取得太陽光伏多步預測之結果進行分析 。最終，透過實驗，使用XGBoost 的集成模型取得研究中最佳的預測結果。24小時後預測之平均絕對百分比誤差為4.83%，168小時後之平均絕對百分比誤差為7.53%。與直接使用原始數據相比，經特徵工程之預測模型實現了 37.5% 的改善。

As awareness of sustainable energy rises, the world is beginning to focus on regulating carbon emissions. Renewable energy is one of the solutions to the current global low-carbon energy target. However, for power plants, the instability of renewable energy will lead to higher generation costs. Microgrids’ ability to accurately forecast solar power generation will help the company plan for power supply and storage, and different forecasting horizons will help at different levels.
However, weather and solar irradiation are the most critical factors affecting solar photovoltaics. The diversity and complexity of weather factors are high, and it is necessary to characterize the weather to find the fundamental factors that affect the prediction. This research uses principal component analysis (PCA), Random Forest, and XGBoost as feature engineering methods and is compared separately. The prediction algorithm is extended by Long Short-term Memory Neural Network (LSTM), including two stacked-LSTM models, Bi-LSTM and AutoEncoder LSTM, which are built and ensembled. This research collects weather data and solar photovoltaic data from two sources. The data are measured hourly, and multi-step predictions are made for 24 hours (1 day) and 168 hours (one week)
ahead.
Finally, a two-stage solar PV multi-step prediction model is proposed. Ultimately, the best prediction results were obtained using the ensemble model of XGBoost. The MAPE after 24 hours was 4.83%, and the MAPE after 168 hours was 7.53%. The feature-engineered prediction model achieved a 37.5% improvement compared to the direct use of original data.

[1] Tsai, Y. C., Chan, Y. K., Ko, F. K., & Yang, J. T. (2018). Integrated operation of renewable energy sources and water resources. Energy Conversion and Management, 160, 439-454.
[2] Wang, F., Zhen, Z., Liu, C., Mi, Z., Hodge, B. M., Shafie-khah, M., & Catalão, J. P. (2018). Image phase shift invariance based cloud motion displacement vector calculation method for ultra-short-term solar PV power forecasting. Energy conversion and management, 157, 123-135.
[3] Wang, F., Xuan, Z., Zhen, Z., Li, K., Wang, T., & Shi, M. (2020). A day-ahead PV power forecasting method based on LSTM-RNN model and time correlation modification under partial daily pattern prediction framework. Energy Conversion and Management, 212, 12766.
[4] Shuvho, M. B. A., Chowdhury, M. A., Ahmed, S., & Kashem, M. A. (2019). Prediction of solar irradiation and performance evaluation of grid connected solar 80KWp PV plant in Bangladesh. Energy Reports, 5, 714-722.
[5] Raza, M. Q., Nadarajah, M., & Ekanayake, C. (2016). On recent advances in PV output power forecast. Solar Energy, 136, 125-144.
[6] Das, U. K., Tey, K. S., Seyedmahmoudian, M., Mekhilef, S., Idris, M. Y. I., Van Deventer, W., ... & Stojcevski, A. (2018). Forecasting of photovoltaic power generation and model optimization: A review. Renewable and Sustainable Energy Reviews, 81, 912-928.
[7] Ahmed, R., Sreeram, V., Mishra, Y., & Arif, M. D. (2020). A review and evaluation of the state-of-the-art in PV solar power forecasting: Techniques and optimization. Renewable and Sustainable Energy Reviews, 124, 109792.
[8] Behera, M. K., Majumder, I., & Nayak, N. (2018). Solar photovoltaic power forecasting using optimized modified extreme learning machine technique. Engineering Science and Technology, an International Journal, 21(3), 428-438.
[9] Ren, Y., Suganthan, P. N., & Srikanth, N. (2015). Ensemble methods for wind and solar power forecasting—A state-of-the-art review. Renewable and Sustainable Energy Reviews, 50, 82-91.
[10] Tsai, Y. C., Chan, Y. K., Ko, F. K., & Yang, J. T. (2018). Integrated operation of renewable energy sources and water resources. Energy Conversion and Management, 160, 439-454.
[11] Wang, F., Zhen, Z., Liu, C., Mi, Z., Hodge, B. M., Shafie-khah, M., & Catalão, J. P. (2018). Image phase shift invariance based cloud motion displacement vector calculation method for ultra-short-term solar PV power forecasting. Energy conversion and management, 157, 123-135.
[12] Wang, F., Xuan, Z., Zhen, Z., Li, K., Wang, T., & Shi, M. (2020). A dayahead PV power forecasting method based on LSTM-RNN model and time correlation modification under partial daily pattern prediction framework. Energy Conversion and Management, 212, 112766.
[13] Shuvho, M. B. A., Chowdhury, M. A., Ahmed, S., & Kashem, M. A. (2019). Prediction of solar irradiation and performance evaluation of grid connected solar 80KWp PV plant in Bangladesh. Energy Reports, 5, 714-722.44
[14] Raza, M. Q., Nadarajah, M., & Ekanayake, C. (2016). On recent advances in PV output power forecast. Solar Energy, 136, 125-144.
[15] Das, U. K., Tey, K. S., Seyedmahmoudian, M., Mekhilef, S., Idris, M. Y. I., Van Deventer, W., ... & Stojcevski, A. (2018). Forecasting of photovoltaic power generation and model optimization: A review. Renewable and Sustainable Energy Reviews, 81, 912-928.
[16] Ahmed, R., Sreeram, V., Mishra, Y., & Arif, M. D. (2020). A review and evaluation of the state-of-the-art in PV solar power forecasting: Techniques and optimization. Renewable and Sustainable Energy Reviews, 124, 109792.
[17] Behera, M. K., Majumder, I., & Nayak, N. (2018). Solar photovoltaic power forecasting using optimized modified extreme learning machine technique. Engineering Science and Technology, an International Journal, 21(3), 428-438.
[18] Ren, Y., Suganthan, P. N., & Srikanth, N. (2015). Ensemble methods for wind and solar power forecasting—A state-of-the-art review. Renewable and Sustainable Energy Reviews, 50, 82-91.
[19] Zdravevski, E., Lameski, P., Mingov, R., Kulakov, A., & Gjorgjevikj, D. (2015, September). Robust histogram-based feature engineering of time series data. In 2015 Federated Conference on Computer Science and Information Systems (FedCSIS) (pp. 381-388). IEEE.
[20] Lin, C. H., Hsieh, W. L., Chen, C. S., Hsu, C. T., & Ku, T. T. (2012). Optimization of photovoltaic penetration in distribution systems considering annual duration curve of solar irradiation. IEEE Transactions on Power Systems, 27(2), 1090-1097.
[21] Cao, L. J., Chua, K. S., Chong, W. K., Lee, H. P., & Gu, Q. M. (2003). A comparison of PCA, KPCA and ICA for dimensionality reduction in support vector machine. Neurocomputing, 55(1-2), 321-336.
[22] Song, F., Guo, Z., & Mei, D. (2010, November). Feature selection using principal component analysis. In 2010 international conference on system science, engineering design and manufacturing informatization (Vol. 1, pp. 27-30). IEEE.
[23] Speiser, J. L., Miller, M. E., Tooze, J., & Ip, E. (2019). A comparison of random forest variable selection methods for classification prediction modeling. Expert systems with applications, 134, 93-101.
[24] Breiman, L. (2001). Random forests machine learning, vol. 45.
[25] Nti, K. O., Adekoya, A., & Weyori, B. (2019). Random forest based feature selection of macroeconomic variables for stock market prediction. American Journal of Applied Sciences, 16(7), 200-212.
[26] Zheng, H., & Wu, Y. (2019). A xgboost model with weather similarity analysis and feature engineering for short-term wind power forecasting. Applied Sciences, 9(15), 3019.
[27] Nguyen, N. Q., Bui, L. D., Van Doan, B., Sanseverino, E. R., Di Cara, D., & Nguyen, Q. D. (2021). A new method for forecasting energy output of a largescale solar power plant based on long short-term memory networks a case study in Vietnam. Electric Power Systems Research, 199, 107427.45
[28] Omer, A. M. (2008). Focus on low carbon technologies: The positive solution. Renewable and Sustainable Energy Reviews, 12(9), 2331-2357.
[29] Hammons, T. J. (2008). Integrating renewable energy sources into European grids. International Journal of Electrical Power & Energy Systems, 30(8), 462-475.
[30] Das, U. K., Tey, K. S., Seyedmahmoudian, M., Mekhilef, S., Idris, M. Y. I., Van Deventer, W., ... & Stojcevski, A. (2018). Forecasting of photovoltaic power generation and model optimization: A review. Renewable and Sustainable Energy Reviews, 81, 912-928.
[31] Qing, X., & Niu, Y. (2018). Hourly day-ahead solar irradiance prediction using weather forecasts by LSTM. Energy, 148, 461-468.
[32] Peng, T., Zhang, C., Zhou, J., & Nazir, M. S. (2021). An integrated framework of Bi-directional long-short term memory (BiLSTM) based on sine cosine algorithm for hourly solar radiation forecasting. Energy, 221, 119887.
[33] Nguyen, H. D., Tran, K. P., Thomassey, S., & Hamad, M. (2021). Forecasting and Anomaly Detection approaches using LSTM and LSTM Autoencoder techniques with the applications in supply chain management. International Journal of Information Management, 57, 102282.
[34] Kuncheva, L. I. (2014). Combining pattern classifiers: methods and algorithms. John Wiley & Sons.
[35] Galicia, A., Talavera-Llames, R., Troncoso, A., Koprinska, I., & MartínezÁlvarez, F. (2019). Multi-step forecasting for big data time series based on ensemble learning. Knowledge-Based Systems, 163, 830-841.
[36] Zhang, B., Zhang, H., Zhao, G., & Lian, J. (2020). Constructing a PM2. 5 concentration prediction model by combining auto-encoder with Bi-LSTM neural networks. Environmental Modelling & Software, 124, 104600.
[37] Liu, H., & Chen, C. (2019). Multi-objective data-ensemble wind speed forecasting model with stacked sparse autoencoder and adaptive decomposition-based error correction. Applied Energy, 254, 113686.
[38] Cheng, H., Ding, X., Zhou, W., & Ding, R. (2019). A hybrid electricity price forecasting model with Bayesian optimization for German energy exchange. International Journal of Electrical Power & Energy Systems, 110, 653-666