臺灣博碩士論文加值系統

近年來電力資源的需求持續增加，電力的供應對於國家經濟發展的至關重要性，電力消耗設備和家電在人們的日常生活中變得更加普遍，進一步提高了生活水平。電力作為由電所提供的能源，在社會和經濟發展中發揮著不可或缺的作用，與可儲存的能源（如煤炭和石油）不同，由於電力的特性，無法大規模儲存，多餘的電力可能造成浪費，而電力短缺則會對社會生產和人民的生活產生不良影響。因此能夠有效進行電力需求預測是非常重要的。由於區間值預測考慮到結果的不確定性，透過建立上下界來預測區間，這樣可以更全面地評估預測的可靠性，而非負矩陣分解(Nonnegative Matrix Factorization, NMF)用於分析數值為零或正數的矩陣，是一個特徵萃取的技術，使用其進行特徵萃取，目的是為了增加預測準確度，最後深度學習算法適用於預測未來並基於當前數據做出決策，因此提出一個結合非負矩陣分解(NMF)與區間值資料的深度學習電力需求預測架構，期待經由非負矩陣分解特徵萃取後有好的結果，並使用另一個特徵萃取技術自動編碼器(Autoencoder, AE)進行比較。本研究使用澳大利亞每30分鐘為單位之實際發電量資料，將區間值資料先進行特徵擴展，再使用非負矩陣分解(NMF)與自動編碼器(AE)進行特徵萃取，並使用深度學習(Deep Learning)方法，包括長短期記憶（Long Short-Term Memory, LSTM）、門控循環單元（Gated Recurrent Unit, GRU）和時間卷積網絡（Temporal Convolution Network, TCN）進行預測。實驗結果顯示，NMF-LSTM、NMF-GRU、NMF-TCN與AE-LSTM、AE-GRU、AE-TCN相比具有較好的預測準確度。

In recent years, the demand for electricity resources has been continuously increasing, making the supply of electricity crucial for national economic development. Electrical consumption devices and household appliances have become more prevalent in people's daily lives, further enhancing the standard of living. Electricity, as an energy source provided by power, plays an indispensable role in social and economic development. Unlike storable energy sources (such as coal and oil), due to the nature of electricity, it cannot be stored on a large scale. Excess electricity may lead to waste, while power shortages can negatively impact social production and people's lives. Therefore, effective electricity demand forecasting is very important. Interval value prediction considers the uncertainty of results by establishing upper and lower bounds to predict the interval, allowing for a more comprehensive assessment of prediction reliability. Nonnegative Matrix Factorization (NMF) is used for analyzing matrices with zero or positive values and is a feature extraction technique. Using it for feature extraction aims to improve prediction accuracy. Finally, deep learning algorithms are suitable for predicting the future and making decisions based on current data. Therefore, a deep learning electricity demand forecasting framework combining NMF and interval value data is proposed, expecting good results through feature extraction. Additionally, another feature extraction technique, Autoencoder (AE), is used for comparison. This study uses actual generation data at 30-minute intervals in Australia. The interval value data is first expanded for feature extraction, then NMF and Autoencoder (AE) are used for feature extraction. Deep learning methods, including Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), and Temporal Convolution Network (TCN), are used for prediction. Experimental results show that NMF-LSTM, NMF-GRU, and NMF-TCN have better prediction accuracy compared to AE-LSTM, AE-GRU, and AE-TCN.

圖次 v
表次 vi
第壹章 緒論 1
第一節 研究背景 1
第二節 研究動機 2
第三節 研究目的 3
第貳章 文獻探討 5
第一節 電力需求 5
第二節 深度學習應用於電力需求之預測 8
第三節 非負矩陣分解 10
第四節 區間值預測 12
第五節 自動編碼器 15
第參章 研究方法 19
第一節 非負矩陣分解(NMF) 19
第二節 長短期記憶網路(LSTM) 22
第三節 門控循環單元（GRU） 26
第四節 時間卷積網路(TCN) 29
第五節 自動編碼器(AutoEncoder) 32
第六節 本研究提出之預測架構 36
第七節 評估準則 43
第肆章 實證結果 45
第一節 資料描述 45
第二節 資料前處理 46
第三節 建立日區間資料 47
第四節 預測結果 54
第伍章 結論與建議 65
第一節 結論 65
第二節 未來研究建議 66
參考文獻 67
附錄 77
附錄一 深度學習與機器學習之比較 77

1.Abedinia, O., & Amjady, N. (2016). Short‐term load forecast of electrical power system by radial basis function neural network and new stochastic search algorithm. International Transactions on Electrical Energy Systems, 26(7), 1511-1525.
2.Abeysingha, A. A. K. U., Sritharan, A. S., Valluvan, R., Ahilan, K., & Jayasinghe, D. H. G. A. E. (2021). Electricity load/demand forecasting in sri lanka using deep learning techniques. In 2021 10th International Conference on Information and Automation for Sustainability (ICIAfS) (pp. 293-298). IEEE.
3.Bai, S., Kolter, J. Z., & Koltun, V. (2018). An empirical evaluation of generic convolutional and recurrent networks for sequence modeling. arXiv preprint arXiv:1803.01271.
4.Bao, W., Li, Q., Xin, L., & Qu, K. (2016). Hyperspectral unmixing algorithm based on nonnegative matrix factorization. In 2016 IEEE international geoscience and remote sensing symposium (IGARSS) (pp. 6982-6985). IEEE.
5.Braun, M. R., Altan, H., & Beck, S. B. M. (2014). Using regression analysis to predict the future energy consumption of a supermarket in the UK. Applied Energy, 130, 305-313.
6.Chatfield, C. (1993). Calculating interval forecasts. Journal of Business & Economic Statistics, 11(2), 121-135.
7.Chen, S., Yang, X., & Li, X. (2022). Research on RF-NMF dimension reduction and CS-LSTM optimized by self-attention mechanism based on sales forecast. In 2022 International Conference on Data Analytics, Computing and Artificial Intelligence (ICDACAI) (pp. 261-272). IEEE.
8.Chou, J. S., Truong, D. N., & Le, T. L. (2020). Interval forecasting of financial time series by accelerated particle swarm-optimized multi-output machine learning system. IEEE Access, 8, 14798-14808.
9.Christoffersen, P. F. (1998). Evaluating interval forecasts. International Economic Review, 841-862.
10.Chung, J., Gulcehre, C., Cho, K., & Bengio, Y. (2014). Empirical evaluation of gated recurrent neural networks on sequence modeling. arXiv preprint arXiv:1412.3555. doi:10.1162/neco_a_01111
11.Da Silva, A. A., & Moulin, L. S. (2000). Confidence intervals for neural network based short-term load forecasting. IEEE Transactions on Power Systems, 15(4), 1191-1196.
12.Ding, Y., Pang, C., Wei, L., Wang, E., Zhao, C., Gao, Q., ... & Li, B. (2024). Short-Term Load Forecasting Method Based on Autoencoder and LSTNet Models. In 2024 IEEE 4th International Conference on Power, Electronics and Computer Applications (ICPECA) (pp. 1057-1062). IEEE.
13.Dol, M., & Geetha, A. (2021). A learning transition from machine learning to deep learning: A survey. In 2021 International Conference on Emerging Techniques in Computational Intelligence (ICETCI) (pp. 89-94). IEEE.
14.Du, G., Zhou, L., Fang, Y., & Yang, M. (2018). Time Series Clustering via NMF in Networks. In 2018 IEEE 16th International Conference on Software Engineering Research, Management and Applications (SERA) (pp. 87-92). IEEE.
15.Ekonomou, L. (2010). Greek long-term energy consumption prediction using artificial neural networks. Energy, 35(2), 512-517.
16.Falcao, G., Alexandre, L. A., Marques, J., Frazão, X., & Maria, J. (2017). On the evaluation of energy-efficient deep learning using stacked autoencoders on mobile GPUs. In 2017 25th Euromicro International Conference on Parallel, Distributed and Network-based Processing (PDP) (pp. 270-273). IEEE.
17.Fumo, N. (2014). A review on the basics of building energy estimation. Renewable and Sustainable Energy Reviews, 31, 53-60.
18.Gaid, M. L., Yousuf, H., Salloum, S. A., & Shaalan, K. (2021). Implementing sequence to sequence neural networks using C#. Net. In The International Conference on Artificial Intelligence and Computer Vision (pp. 742-753). Cham: Springer International Publishing.
19.Gers, F. A., & Schmidhuber, J. (2000). Recurrent nets that time and count. In Proceedings of the IEEE-INNS-ENNS International Joint Conference on Neural Networks. IJCNN 2000. Neural Computing: New Challenges and Perspectives for the New Millennium (Vol. 3, pp. 189-194). IEEE.
20.Ghimire, S., Nguyen-Huy, T., AL-Musaylh, M. S., Deo, R. C., Casillas-Pérez, D., & Salcedo-Sanz, S. (2023). A novel approach based on integration of convolutional neural networks and echo state network for daily electricity demand prediction. Energy, 275, 127430.
21.Gligorić, Z., Savić, S. Š., Grujić, A., Negovanović, M., & Musić, O. (2018). Short-term electricity price forecasting model using interval-valued autoregressive process. Energies, 11(7), 1911.
22.Guo, Q., Feng, Y., Sun, X., & Zhang, L. (2017). Power demand forecasting and application based on SVR. Procedia Computer Science, 122, 269-275.
23.Hajek, P., Prochazka, O., & Froelich, W. (2018). Interval-valued intuitionistic fuzzy cognitive maps for stock index forecasting. In 2018 IEEE Conference on Evolving and Adaptive Intelligent Systems (EAIS) (pp. 1-7). IEEE.
24.Hiba, C., Tarek, K. M., & Belkacem, C. (2020). Stacked Denoising Autoencoder network for short-term prediction of electrical Algerian load. In 2020 7th International Conference on Control, Decision and Information Technologies (CoDIT) (Vol. 1, pp. 189-194). IEEE.
25.Hochreiter, S., & Schmidhuber, J. (1997). Long short-term memory. Neural computation, 9(8), 1735-1780.
26.Hsu, H. L., & Wu, B. (2008). Evaluating forecasting performance for interval data. Computers & Mathematics with Applications, 56(9), 2155-2163. doi: 10.1016/j.camwa.2008.03.042
27.Hu, H., Wang, L., Peng, L., & Zeng, Y. R. (2020). Effective energy consumption forecasting using enhanced bagged echo state network. Energy, 193, 116778.
28.Hu, Z., Cai, J., Wang, Z., & Qin, F. (2022). Minimum Sample Size Estimation Method of Electromagnetic Effect Test Based on Confidence Interval. In 2022 IEEE 5th International Conference on Electronics Technology (ICET) (pp. 249-254). IEEE.
29.Irankhah, A., Rezazadeh, S., Moghaddam, M. H. Y., & Ershadi-Nasab, S. (2021). Hybrid deep learning method based on lstm-autoencoder network for household short-term load forecasting. In 2021 7th International Conference on Signal Processing and Intelligent Systems (ICSPIS) (pp. 1-6). IEEE.
30.Kale, R. V., & Pohekar, S. D. (2013). Long-range forecasting of electricity demand and supply for Maharashtra. In 2013 International Conference on Renewable Energy and Sustainable Energy (ICRESE) (pp. 5-8). IEEE.
31.Kaneko, N., Iwabuchi, K., Kato, K., Watari, D., Zhao, D., Taniguchi, I., ... & Onoye, T. (2022, October). An Evaluation of Electricity Demand Forecasting Models for Smart Energy Management Systems. In 2022 19th International SoC Design Conference (ISOCC) (pp. 195-196). IEEE.
32.Kang, T., Lim, D. Y., Tayara, H., & Chong, K. T. (2020). Forecasting of power demands using deep learning. Applied Sciences, 10(20), 7241.
33.Khashei, M., Montazeri, M. A., & Bijari, M. (2015). Comparison of four interval ARIMA-base time series methods for exchange rate forecasting. International Journal of Mathematical Sciences and Computing, 1(1), 21-34.
34.Kim, J. H., Wong, K., Athanasopoulos, G., & Liu, S. (2011). Beyond point forecasting: Evaluation of alternative prediction intervals for tourist arrivals. International Journal of Forecasting, 27(3), 887-901.
35.Konstantinou, T., Savvopoulos, N., & Hatziargyriou, N. (2021). Scenario Based Probabilistic Energy Demand Forecasting using Autoencoders and Gaussian Mixture Models. In 2021 International Conference on Smart Energy Systems and Technologies (SEST) (pp. 1-6). IEEE
36.Kumar, R., Aggarwal, R. K., & Sharma, J. D. (2013). Energy analysis of a building using artificial neural network: A review. Energy and Buildings, 65, 352-358.
37.Kumru, M., & Kumru, P. Y. (2015). Calendar-based short-term forecasting of daily average electricity demand. In 2015 International Conference on Industrial Engineering and Operations Management (IEOM) (pp. 1-5). IEEE.
38.Lee, D. D., & Seung, H. S. (1999). Learning the parts of objects by non-negative matrix factorization. Nature, 401(6755), 788-791.
39.Li, C., Guo, Q., Shao, L., Li, J., & Wu, H. (2022a). Research on short-term load forecasting based on optimized gru neural network. Electronics, 11(22), 3834.
40.Li, H., Li, P., Zhang, Y., Zu, W., & Zheng, Y. (2022b). Research on electricity demand forecasting based on LSTM-SVR. In 2022 2nd International Conference on Algorithms, High Performance Computing and Artificial Intelligence (AHPCAI) (pp. 233-239). IEEE.
41.Li, X., & Tyagi, A. (2023). Block-Active ADMM to Minimize NMF with Bregman Divergences. Sensors, 23(16), 7229.
42.Li, X., Bao, C., & Cui, Z. (2021). An NMF-based MMSE Approach for Single Channel Speech Enhancement Using Densely Connected Convolutional Network. In 2021 IEEE International Conference on Signal Processing, Communications and Computing (ICSPCC) (pp. 1-5). IEEE.
43.Liu, F., Cai, M., Wang, L., & Lu, Y. (2019). An ensemble model based on adaptive noise reducer and over-fitting prevention LSTM for multivariate time series forecasting. IEEE Access, 7, 26102-26115.
44.Liu, G., Xiao, F., Lin, C. T., & Cao, Z. (2020). A fuzzy interval time-series energy and financial forecasting model using network-based multiple time-frequency spaces and the induced-ordered weighted averaging aggregation operation. IEEE Transactions on Fuzzy Systems, 28(11), 2677-2690.
45.Ma, C., Liang, L., Chen, Y., & Zhang, Q. (2020). Feature Extraction for Fault Diagnosis of Machine based on Kernel Nonnegative Matrix Factorization. In 2020 IEEE 4th Information Technology, Networking, Electronic and Automation Control Conference (ITNEC) (Vol. 1, pp. 1412-1416). IEEE.
46.Ma, Z., Prljaca, Z., & Jørgensen, B. N. (2016). The international electricity market infrastructure-insight from the nordic electricity market. In 2016 13th International Conference on the European Energy Market (EEM) (pp. 1-5). IEEE.
47.Maciel, L. (2023). A trading strategy based on BitCoin high and low prices: the role of an evolving fuzzy model for interval-valued time series forecasting. In 2023 IEEE International Conference on Fuzzy Systems (FUZZ) (pp. 1-6). IEEE.
48.Maciel, L., Gomide, F., & Ballini, R. (2015). Evolving possibilistic fuzzy modeling for financial interval time series forecasting. In 2015 Annual Conference of the North American Fuzzy Information Processing Society (NAFIPS) held jointly with 2015 5th World Conference on Soft Computing (WConSC) (pp. 1-6). IEEE.
49.Maia, A. L. S., de Carvalho, F. D. A., & Ludermir, T. B. (2008). Forecasting models for interval-valued time series. Neurocomputing, 71(16-18), 3344-3352.
50.Mansouri, N., & Lachiri, Z. (2020). Laughter synthesis: A comparison between Variational autoencoder and Autoencoder. In 2020 5th International Conference on Advanced Technologies for Signal and Image Processing (ATSIP) (pp. 1-6). IEEE.
51.Mardira, L., Saha, T. K., & Eghbal, M. (2014). Investigating impacts of battery energy storage systems on electricity demand profile. In 2014 Australasian Universities Power Engineering Conference (AUPEC) (pp. 1-5). IEEE.
52.Mei, J., De Castro, Y., Goude, Y., Azaïs, J. M., & Hébrail, G. (2018). Nonnegative matrix factorization with side information for time series recovery and prediction. IEEE Transactions on Knowledge and Data Engineering, 31(3), 493-506.
53.Mei, X., Xu, C., Liu, L., & Yang, Y. (2015). Learning part-based dictionaries by NMF for crude oil market prediction. In 2015 4th International Conference on Computer Science and Network Technology (ICCSNT) (Vol. 1, pp. 624-628). IEEE.
54.Mezaache, H., & Bouzgou, H. (2018). Auto-encoder with neural networks for wind speed forecasting. In 2018 International Conference on Communications and Electrical Engineering (ICCEE) (pp. 1-5). IEEE.
55.Michalski, R. S., Carbonell, J. G., & Mitchell, T. M. (Eds.). (2013). Machine learning: An artificial intelligence approach. Springer Science & Business Media.
56.Mo, R., Pei, Y., Venkatarayalu, N., Nathaniel, P., Premkumar, A. B., & Sun, S. (2021). An unsupervised TCN-based outlier detection for time series with seasonality and trend. In 2021 IEEE VTS 17th Asia Pacific Wireless Communications Symposium (APWCS) (pp. 1-5). IEEE.
57.Morita, H., Kase, T., Tamura, Y., & Iwamoto, S. (1996). Interval prediction of annual maximum demand using grey dynamic model. International Journal of Electrical Power & Energy Systems, 18(7), 409-413.
58.Mutinda, F., Nakashima, A., Takeuchi, K., Sasaki, Y., & Onizuka, M. (2019). Time series link prediction using nmf. Journal of Information Processing, 27, 752-761.
59.Najafi, A., Homaee, O., Golshan, M., Jasinski, M., & Leonowicz, Z. (2023). Application of extreme learning machine-autoencoder to medium term electricity price forecasting. IEEE Transactions on Industry Applications.
60.Nasrabadi, N. M. (2007). Pattern recognition and machine learning. Journal of Electronic Imaging, 16(4), 049901.
61.Pekaslan, D., Wagner, C., Garibaldi, J. M., Marin, L. G., & Sáez, D. (2020). Uncertainty-aware forecasting of renewable energy sources. In 2020 IEEE International Conference on Big Data and Smart Computing (BigComp) (pp. 240-246). IEEE.
62.Rana, M., Koprinska, I., & Agelidis, V. G. (2015). 2D-interval forecasts for solar power production. Solar Energy, 122, 191-203.
63.Rimal, R., Brannon, M., Wang, Y., & Yang, X. (2022). Comparative study of machine learning and deep learning methods on ASD classification. arXiv preprint arXiv:2209.08601.
64.Roque, A. M. S., Maté, C., Arroyo, J., & Sarabia, Á. (2007). iMLP: Applying multi-layer perceptrons to interval-valued data. Neural Processing Letters, 25, 157-169.
65.Sang, S., & Li, L. (2024). A Novel Variant of LSTM Stock Prediction Method Incorporating Attention Mechanism. Mathematics, 12(7), 945.
66.Singh, S., & Tripathi, M. M. (2021). A Comparative Analysis of Extreme Gradient Boosting Technique with Long Short-Term Memory and Layered Recurrent Neural Network for Electricity Demand Forecas. In 2021 International Conference on Recent Trends on Electronics, Information, Communication & Technology (RTEICT) (pp. 297-302). IEEE.
67.Son, H., & Kim, C. (2020). A deep learning approach to forecasting monthly demand for residential–sector electricity. Sustainability, 12(8), 3103.
68.Suresh, V., Aksan, F., Janik, P., Sikorski, T., & Revathi, B. S. (2022). Probabilistic LSTM-Autoencoder based hour-ahead solar power forecasting model for intra-day electricity market participation: A Polish case study. IEEE Access, 10, 110628-110638.
69.Wang, D., Zhang, R., & Zhao, L. (2022). A Multivariate Load Prediction Method of Integrated Energy Systems Based on MMoE-TCN. In 2022 7th International Conference on Power and Renewable Energy (ICPRE) (pp. 596-602). IEEE.
70.Wang, H., Wang, L., & Ma, L. (2021b). Anomaly detection of hydropower bearing based on convolutional neural network autoencoder. In 2021 IEEE 4th International Conference on Computer and Communication Engineering Technology (CCET) (pp. 430-433). IEEE.
71.Wang, J., Tang, G., & Wang, Y. (2023). Application in Student Performance Prediction Using Graph Regularization Nonnegative Matrix Factorization. In 2023 10th International Conference on Dependable Systems and Their Applications (DSA) (pp. 816-820). IEEE.
72.Wang, S. (2023). Application of Nonnegative Matrix Factorization in Intelligent Patrol Inspection of Substations. In 2023 4th International Conference on Big Data, Artificial Intelligence and Internet of Things Engineering (ICBAIE) (pp. 255-258). IEEE.
73.Wang, S., & Xu, X. (2006). Simplified building model for transient thermal performance estimation using GA-based parameter identification. International Journal of Thermal Sciences, 45(4), 419-432.
74.Wang, Y., Sun, S., Chen, X., Zeng, X., Kong, Y., Chen, J., ... & Wang, T. (2021a). Short-term load forecasting of industrial customers based on SVMD and XGBoost. International Journal of Electrical Power & Energy Systems, 129, 106830.
75.Xiaozhi, L., Yang, G., & Yinghua, Y. (2020). Fault Diagnosis Based on Batch-normalized Stacked Sparse Autoencoder. In 2020 39th Chinese Control Conference (CCC) (pp. 4141-4146). IEEE.
76.Xu, M., Dong, Y., Li, Z., Han, M., & Xing, T. (2018). A novel time series prediction model based on deep sparse autoencoder. In 2018 37th Chinese Control Conference (CCC) (pp. 1678-1682). IEEE.
77.Xu, Z., Kang, Y., Cao, Y., & Yue, L. (2018). Residual autoencoder-LSTM for city region vehicle emission pollution prediction. In 2018 IEEE 14th International Conference on Control and Automation (ICCA) (pp. 811-816). IEEE.
78.Yamada, K., & Mori, H. (2023). Practical Application of Deep Modified Autoencoder Technique to Electricity Price Forecasting. In 2023 IEEE International Conference on Energy Technologies for Future Grids (ETFG) (pp. 1-7). IEEE.
79.Yang, G., Du, S., Duan, Q., & Su, J. (2022). Short-term demand forecasting method in power markets based on the KSVM–TCN–GBRT. Computational Intelligence and Neuroscience, 2022.
80.Ye, Q., Chen, W., Zhu, L., Lin, H., Zhang, F., Tan, J., ... & Zhao, Y. (2023). Data-driven Approaches Predict Hourly Electricity Demand Profiles at Industry and City-level. In 2023 8th Asia Conference on Power and Electrical Engineering (ACPEE) (pp. 905-910). IEEE.
81.Yuan, Y., Feng, Y., & Lu, X. (2015). Projection-based NMF for hyperspectral unmixing. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 8(6), 2632-2643.
82.Zahedi, G., Azizi, S., Bahadori, A., Elkamel, A., & Alwi, S. R. W. (2013). Electricity demand estimation using an adaptive neuro-fuzzy network: a case study from the Ontario province–Canada. Energy, 49, 323-328.
83.Zhang, H., Liang, Q., Wang, R., & Wu, Q. (2020). Stacked model with autoencoder for financial time series prediction. In 2020 15th International Conference on Computer Science & Education (ICCSE) (pp. 222-226). IEEE.
84.Zhang, Y., Jing, R., Zhang, X., Hu, R., Wang, N., Li, L., & Kang, Y. (2023a). A study on Short-Term Electricity Load Forecasting for Industrial Parks method using QPSO-TCN Based on Autoencoder. In 2023 2nd International Conference on Smart Grids and Energy Systems (SGES) (pp. 180-187). IEEE.
85.Zhang, Y., Mu, S., Chen, C., Pei, J., & Han, J. (2023b). Weekly Electricity Forecasting Method Based on Double-Layer lightGBM-GRU-IBES Algorithm. In 2023 IEEE International Conference on Image Processing and Computer Applications (ICIPCA) (pp. 1784-1792). IEEE.
86.Zhu, J., Hu, W., Xu, X., Luo, S., Liu, H., Hu, C., ... & Huang, Q. (2022). The Research on the Construction of Confidence Interval Model for Solar, Hydropower and Load Demand. In 2022 4th Asia Energy and Electrical Engineering Symposium (AEEES) (pp. 484-491). IEEE.