Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
110 tokens/sec
GPT-4o
56 tokens/sec
Gemini 2.5 Pro Pro
44 tokens/sec
o3 Pro
6 tokens/sec
GPT-4.1 Pro
47 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

A comparative assessment of deep learning models for day-ahead load forecasting: Investigating key accuracy drivers (2302.12168v2)

Published 23 Feb 2023 in cs.LG and cs.AI

Abstract: Short-term load forecasting (STLF) is vital for the effective and economic operation of power grids and energy markets. However, the non-linearity and non-stationarity of electricity demand as well as its dependency on various external factors renders STLF a challenging task. To that end, several deep learning models have been proposed in the literature for STLF, reporting promising results. In order to evaluate the accuracy of said models in day-ahead forecasting settings, in this paper we focus on the national net aggregated STLF of Portugal and conduct a comparative study considering a set of indicative, well-established deep autoregressive models, namely multi-layer perceptrons (MLP), long short-term memory networks (LSTM), neural basis expansion coefficient analysis (N-BEATS), temporal convolutional networks (TCN), and temporal fusion transformers (TFT). Moreover, we identify factors that significantly affect the demand and investigate their impact on the accuracy of each model. Our results suggest that N-BEATS consistently outperforms the rest of the examined models. MLP follows, providing further evidence towards the use of feed-forward networks over relatively more sophisticated architectures. Finally, certain calendar and weather features like the hour of the day and the temperature are identified as key accuracy drivers, providing insights regarding the forecasting approach that should be used per case.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (75)
  1. Optuna: A Next-generation Hyperparameter Optimization Framework, in: Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, Association for Computing Machinery. pp. 2623–2631. doi:10.1145/3292500.3330701.
  2. Beginning MLOps with MLFlow. Apress. doi:10.1007/978-1-4842-6549-9.
  3. Enhanced Short-Term Load Forecasting Using Artificial Neural Networks. Energies 14, 7788. doi:10.3390/en14227788.
  4. Deep Reinforcement Learning for Demand Response in Distribution Networks. IEEE Transactions on Smart Grid 12, 1496–1506. doi:10.1109/TSG.2020.3037066.
  5. An Empirical Evaluation of Generic Convolutional and Recurrent Networks for Sequence Modeling. arXiv:1803.01271 .
  6. Algorithms for Hyper-Parameter Optimization. Advances in Neural Information Processing Systems 24. doi:https://dl.acm.org/doi/10.5555/2986459.2986743.
  7. Group-based chaos genetic algorithm and non-linear ensemble of neural networks for short-term load forecasting. IET Generation, Transmission and Distribution 10, 1440–1447. doi:10.1049/iet-gtd.2015.1068.
  8. Robust locally weighted regression and smoothing scatterplots. Journal of the American Statistical Association 74, 829–836. doi:10.1080/01621459.1979.10481038.
  9. Short-Term City Electric Load Forecasting with Considering Temperature Effects: An Improved ARIMAX Model. Mathematical Problems in Engineering 2015. doi:10.1155/2015/589374.
  10. Short-Term Load Forecasting with Neural Network Ensembles: A Comparative Study [Application Notes]. IEEE Computational Intelligence Magazine 6, 47–56. doi:10.1109/MCI.2011.941590.
  11. On short-term load forecasting using machine learning techniques and a novel parallel deep LSTM-CNN approach. IEEE Access 9, 31191–31212. doi:10.1109/ACCESS.2021.3060290.
  12. Deep learning for time series forecasting: The electric load case. CAAI Transactions on Intelligence Technology 7, 1–25. doi:10.1049/CIT2.12060.
  13. A Global Modeling Approach for Load Forecasting in Distribution Networks [Preprint]. arXiv:2204.00493 .
  14. Temporal Convolutional Network Based Short-term Load Forecasting Model, in: 2020 IEEE 9th Data Driven Control and Learning Systems Conference (DDCLS), IEEE. pp. 584–589. doi:10.1109/DDCLS49620.2020.9275148.
  15. Short term load forecasting and the effect of temperature at the low voltage level. International Journal of Forecasting 35, 1469–1484. doi:10.1016/J.IJFORECAST.2018.10.007.
  16. Methods and Models for Electric Load Forecasting: A Comprehensive Review. Logistics & Sustainable Transport 11, 51–76. doi:10.2478/JLST-2020-0004.
  17. Short-term load forecasting using an attended sequential encoder-stacked decoder model with online training. Applied Sciences (Switzerland) 11. doi:10.3390/app11114927.
  18. A multi-agent system architecture for smart grid management and forecasting of energy demand in virtual power plants. IEEE Communications Magazine 51, 106–113. doi:10.1109/MCOM.2013.6400446.
  19. A survey on electric power demand forecasting: Future trends in smart grids, microgrids and smart buildings. IEEE Communications Surveys and Tutorials 16, 1460–1495. doi:10.1109/SURV.2014.032014.00094.
  20. Darts: User-Friendly Modern Machine Learning for Time Series. Journal of Machine Learning Research 23, 1–6. doi:10.48550/arxiv.2110.03224.
  21. Short term load forecasting using a multilayer neural network with an adaptive learning algorithm. IEEE Transactions on Power Systems 7, 141–149. doi:10.1109/59.141697.
  22. Analysis of the value for unit commitment of improved load forecasts. IEEE Transactions on Power Systems 14, 1342–1348. doi:10.1109/59.801894.
  23. Long Short-Term Memory. Neural Computation doi:10.1162/neco.1997.9.8.1735.
  24. Short-Term Electricity Load Forecasting Based on Temporal Fusion Transformer Model. IEEE Access 10, 106296–106304. doi:10.1109/ACCESS.2022.3211941.
  25. Another look at measures of forecast accuracy. International Journal of Forecasting 22, 679–688. doi:10.1016/j.ijforecast.2006.03.001.
  26. An efficient approach for short term load forecasting using artificial neural networks. International Journal of Electrical Power & Energy Systems 28, 525–530. doi:10.1016/j.ijepes.2006.02.014.
  27. Visualising forecasting algorithm performance using time series instance spaces. International Journal of Forecasting 33, 345–358. doi:10.1016/J.IJFORECAST.2016.09.004.
  28. Artificial Intelligence for Next Generation Energy Services Across Europe - The I-NERGY Project, in: ES 2021 : 19th International Conference e-Society 2021, Lisbon. pp. 61–68.
  29. The M3 competition: Statistical tests of the results. International Journal of Forecasting 21, 397–409. doi:10.1016/J.IJFORECAST.2004.10.003.
  30. Neural network ensemble operators for time series forecasting. Expert Systems with Applications 41, 4235–4244. doi:10.1016/J.ESWA.2013.12.011.
  31. Application of a fuzzy neural network combined with a chaos genetic algorithm and simulated annealing to short-term load forecasting. IEEE Transactions on Evolutionary Computation 10, 330–340. doi:10.1109/TEVC.2005.857075.
  32. Temporal Fusion Transformers for Interpretable Multi-horizon Time Series Forecasting. International Journal of Forecasting 37, 1748–1764. URL: https://arxiv.org/abs/1912.09363v3, doi:10.1016/j.ijforecast.2021.03.012.
  33. Temporal Fusion Transformers for interpretable multi-horizon time series forecasting. International Journal of Forecasting 37, 1748–1764. doi:10.1016/J.IJFORECAST.2021.03.012.
  34. A survey of transformers. AI Open 3, 111–132. doi:10.1016/J.AIOPEN.2022.10.001.
  35. Short term load forecasting using neural network trained with genetic algorithm & particle swarm optimization, in: Proceedings - 1st International Conference on Emerging Trends in Engineering and Technology, ICETET 2008, pp. 606–611. doi:10.1109/ICETET.2008.94.
  36. A linear regression-based study for temperature sensitivity analysis of iran electrical load, in: Proceedings of the IEEE International Conference on Industrial Technology. doi:10.1109/ICIT.2008.4608590.
  37. FFORMA: Feature-based forecast model averaging. International Journal of Forecasting 36, 86–92. doi:10.1016/J.IJFORECAST.2019.02.011.
  38. Toward explainable electrical load forecasting of buildings: A comparative study of tree-based ensemble methods with Shapley values. Sustainable Energy Technologies and Assessments 54, 102888. doi:10.1016/J.SETA.2022.102888.
  39. Long-term load forecasting: models based on MARS, ANN and LR methods. Central European Journal of Operations Research 27, 1033–1049. doi:10.1007/S10100-018-0531-1/FIGURES/9.
  40. N-BEATS: Neural basis expansion analysis for interpretable time series forecasting. arXiv doi:https://doi.org/10.48550/arXiv.1905.10437.
  41. N-BEATS neural network for mid-term electricity load forecasting. Applied Energy 293, 116918. doi:10.1016/J.APENERGY.2021.116918.
  42. Economic impact assessment of load forecast errors considering the cost of interruptions. 2006 IEEE Power Engineering Society General Meeting, PES doi:10.1109/PES.2006.1709231.
  43. An Introduction to Convolutional Neural Networks. arXiv:1511.08458 .
  44. MATRYCS A Big Data Architecture for Advanced Services in the Building Domain. Energies 15, 2568. doi:10.3390/EN15072568.
  45. Scikit-learn: Machine Learning in Python. Journal of Machine Learning Research 12, 2825–2830.
  46. In Search of Deep Learning Architectures for Load Forecasting: A Comparative Analysis and the Impact of the Covid-19 Pandemic on Model Performance, in: 2022 13th International Conference on Information, Intelligence, Systems and Applications (IISA), IEEE. pp. 1–8. doi:10.1109/IISA56318.2022.9904363.
  47. Targeted demand response for flexible energy communities using clustering techniques. Sustainable Energy, Grids and Networks 36, 101134. URL: https://linkinghub.elsevier.com/retrieve/pii/S235246772300142X, doi:10.1016/J.SEGAN.2023.101134.
  48. ‘Horses for Courses’ in demand forecasting. European Journal of Operational Research 237, 152–163. doi:10.1016/J.EJOR.2014.02.036.
  49. Factors affecting electricity demand in Athens, Greece and London, UK: A comparative assessment. Energy 34, 1855–1863. doi:10.1016/J.ENERGY.2009.07.033.
  50. A short-term load forecasting method using integrated CNN and LSTM network. IEEE Access 9, 32436–32448. doi:10.1109/ACCESS.2021.3060654.
  51. R&D Nester, 2023. R&D Nester | Homepage. URL: https://www.rdnester.pt/en-GB.
  52. REPRS, 2013. Regulation for Energy Performance of Residential Buildings. URL: https://diariodarepublica.pt/dr/detalhe/decreto-lei/118-2013-499237.
  53. Climatologia e Anos Meteorológicos de Referência para o Sistema Nacional de Certificação de Edifícios , 55.
  54. The Shapley Value in Machine Learning. IJCAI International Joint Conference on Artificial Intelligence , 5572–5579doi:10.48550/arxiv.2202.05594.
  55. rp5, 2023a. Weather archive in Lisbon. URL: https://rp5.ru/Weather_archive_in_Lisbon.
  56. rp5, 2023b. Weather archive in Lisbon / Portela (airport), METAR. URL: https://rp5.ru/Weather_archive_in_Lisbon,_Portela_(airport),_METAR.
  57. Short-term load forecasting using encoder-decoder wavenet: Application to the french grid. Energies 14. doi:10.3390/en14092524.
  58. A Novel CNN-GRU-Based Hybrid Approach for Short-Term Residential Load Forecasting. IEEE Access 8, 143759–143768. doi:10.1109/ACCESS.2020.3009537.
  59. Deep Learning for Load Forecasting: Sequence to Sequence Recurrent Neural Networks with Attention. IEEE Access 8, 36411–36426. doi:10.1109/ACCESS.2020.2975738.
  60. Short-Term Forecasting in Smart Electric Grid Using N-BEATS. ICPC2T 2022 - 2nd International Conference on Power, Control and Computing Technologies, Proceedings doi:10.1109/ICPC2T53885.2022.9776757.
  61. Forecasting week-ahead hourly electricity prices in Belgium with statistical and machine learning methods. Mathematical Modelling of Contemporary Electricity Markets , 59–74doi:10.1016/B978-0-12-821838-9.00005-0.
  62. Are forecasting competitions data representative of the reality? International Journal of Forecasting 36, 37–53. doi:10.1016/J.IJFORECAST.2018.12.007.
  63. A historical turning point? Early evidence on how the Russia-Ukraine war changes public support for clean energy policies. Energy Research & Social Science 91, 102758. doi:10.1016/J.ERSS.2022.102758.
  64. Stimmel Carol, 2019. Big data analytics strategies for the smart grid. Auerbach Publications.
  65. FFORMPP: Feature-based forecast model performance prediction. International Journal of Forecasting 38, 920–943. doi:10.1016/J.IJFORECAST.2021.07.002.
  66. Short-Term Load Forecasting Using Channel and Temporal Attention Based Temporal Convolutional Network. Electric Power Systems Research 205, 107761. doi:10.1016/J.EPSR.2021.107761.
  67. UCI, 2022. UCI Machine Learning Repository | ElectricityLoadDiagrams20112014 Data Set. URL: https://archive.ics.uci.edu/ml/datasets/ElectricityLoadDiagrams20112014.
  68. Attention Is All You Need, in: Advances in Neural Information Processing Systems, Neural information processing systems foundation. pp. 5999–6009.
  69. Energy flexibility prediction for data center engagement in demand response programs. Sustainability 12, 1417. doi:https://doi.org/10.3390/su12041417.
  70. The BD4NRG Reference Architecture for Big Data Driven Energy Applications. 13th International Conference on Information, Intelligence, Systems and Applications, IISA 2022 doi:10.1109/IISA56318.2022.9904424.
  71. Probabilistic Load Forecasting via Neural Basis Expansion Model Based Prediction Intervals. IEEE Transactions on Smart Grid 12, 3648–3660. doi:10.1109/TSG.2021.3066567.
  72. An explainable framework for load forecasting of a regional integrated energy system based on coupled features and multi-task learning. Protection and Control of Modern Power Systems 7, 1–14. doi:10.1186/S41601-022-00245-Y/TABLES/7.
  73. Multi-temporal-spatial-scale temporal convolution network for short-term load forecasting of power systems. Applied Energy 283, 116328. doi:10.1016/J.APENERGY.2020.116328.
  74. Short-Term Electrical Load Forecasting Based on Time Augmented Transformer. International Journal of Computational Intelligence Systems 15, 1–11. doi:10.1007/S44196-022-00128-Y/FIGURES/6.
  75. Short-term load forecasting of Australian national electricity market by an ensemble model of extreme learning machine. IET Generation, Transmission and Distribution 7, 391–397. doi:10.1049/iet-gtd.2012.0541.
User Edit Pencil Streamline Icon: https://streamlinehq.com
Authors (7)
Citations (12)
View on Semantic Scholar

Summary

We haven't generated a summary for this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize Now