Electricity consumption prediction using artificial intelligence. (English) Zbl 07722447
Summary: The measurement of electricity consumption at 15-minute granularity, including for households, is increasingly mandated in the EU and this also allows, once sufficient data have been collected, the prediction of future consumption at the same time intervals.
In this paper, we present preliminary results of the industry project that aims to build AI models for next-day electricity consumption at 15-minute granularity. We have identified the main influencing factors, developed scripts and databases to collect data about these features and about the past electricity consumption at 15-minute granularity for each measuring point, and, finally, developed three AI models to predict the future electricity consumption for each 15-minute interval and each measurement point.
We provide descriptive analyses for all measuring points that were in the database in April 2022 and show that for computing the prediction of accumulated electricity consumption at 15-minute granularity, it is much more accurate (in terms of mean absolute percentage error – MAPE) to compute the prediction for each measuring point and accumulate these predictions. An evaluation of the models on the list of the 10 outstanding measuring points (according to the data provider) shows that our predictions achieve very good MAPE. Additionally, we have provided an evaluation of possible ways of parallelization within R, and laid out results of a computational study using parallel, doParallel, and foreach R libraries.
In this paper, we present preliminary results of the industry project that aims to build AI models for next-day electricity consumption at 15-minute granularity. We have identified the main influencing factors, developed scripts and databases to collect data about these features and about the past electricity consumption at 15-minute granularity for each measuring point, and, finally, developed three AI models to predict the future electricity consumption for each 15-minute interval and each measurement point.
We provide descriptive analyses for all measuring points that were in the database in April 2022 and show that for computing the prediction of accumulated electricity consumption at 15-minute granularity, it is much more accurate (in terms of mean absolute percentage error – MAPE) to compute the prediction for each measuring point and accumulate these predictions. An evaluation of the models on the list of the 10 outstanding measuring points (according to the data provider) shows that our predictions achieve very good MAPE. Additionally, we have provided an evaluation of possible ways of parallelization within R, and laid out results of a computational study using parallel, doParallel, and foreach R libraries.
MSC:
| 90Bxx | Operations research and management science |
Keywords:
electricity demand; load forecasting; neural network; random forest; accuracy; parallelizationReferences:
| [1] | Berk, RA, Random forests, Statistical learning from a regression perspective (2020), Cham: Springer, Cham · Zbl 1435.62001 · doi:10.1007/978-3-030-40189-4_5 |
| [2] | Chodorow C (2010) Introduction to MongoDB. Free and Open Source Software Developers’ European Meeting (FOSDEM) |
| [3] | Čegovnik T, Dobrovoljc A, Povh J, Rogar M (2021) : Electricity consumption prediction using artificial intelligence. V: DROBNE, Samo (ur.), SOR ‘21 proceedings: the 16th International Symposium on Operational Research in Slovenia : September 22-24, 2021, online. Ljubljana: Slovenian Society Informatika, Section for Operational Research, pp. 181-187 · Zbl 07722447 |
| [4] | De’Ath, G., Multivariate regression trees: a new technique for modeling species-environment relationships, Ecology, 83, 1105-1117 (2002) |
| [5] | Genuer, R.; Poggi, JM, Random forests, Random forests with R. Use R! (2020), Cham: Springer, Cham · Zbl 1448.62004 · doi:10.1007/978-3-030-56485-8_3 |
| [6] | Hastie T, Tibshirani R, Friedman J (2009) : “The elements of statistical learning: data mining, inference, and prediction”. Springer Science & Business Media, 2009 · Zbl 1273.62005 |
| [7] | Kapustina, E.; Shutov, E.; Barskaya, A.; Kalganova, A., Predicting Electric Energy Consumption for a Jerky Enterprise, Energy and Power Engineering, 12, 396-406 (2020) · doi:10.4236/epe.2020.126024 |
| [8] | Kim JY, Cho SB (2021) : Interpretable Deep Learning with Hybrid Autoencoders to Predict Electric Energy Consumption. In: Herrero Á., Cambra C., Urda D., Sedano J., Quintián H., Corchado E. (eds) 15th International Conference on Soft Computing Models in Industrial and Environmental Applications (SOCO 2020). SOCO 2020. Advances in Intelligent Systems and Computing, vol 1268. Springer, Cham. doi:10.1007/978-3-030-57802-2_13 |
| [9] | Banker, K., MongoDB in Action (2011), Greenwich, CT, USA: Manning Publications Co, Greenwich, CT, USA |
| [10] | Polimis, Confidence intervals for Random forests in Python, J Open Source Softw, 2, 19, 124 (2017) · doi:10.21105/joss.00124 |
| [11] | Wei R, Wang J, Gan Q, Dang X, Wang H (2019) : Predicting Electricity Usage Based on Deep Neural Network*, 2019 IEEE International Conference on Computational Intelligence and Virtual Environments for Measurement Systems and Applications (CIVEMSA), Tianjin, China, pp. 1-6, doi: doi:10.1109/CIVEMSA45640.2019.9071602 |
| [12] | Wei-ping Z, Ming-xin L, Huan C (2011) : Using MongoDB to implement textbook management system instead of MySQL. In 2011 IEEE 3rd International Conference on Communication Software and Networks, pp. 303-305. IEEE |
| [13] | Chen Y, Guo M, Chen Z, Chen Z, Ji Y (2022) : Physical energy and data-driven models in building energy prediction: A review. Energy Reports 8:2656-2671. Elsevier |
| [14] | Wang Z, Hong T, Piette MA (2020) : Building thermal load prediction through shallow machine learning and deep learning. Energy Technologies Area April 2020 |
| [15] | Amasyali K, El-Gohary NM (2016) : A review of data-driven building energy consumption prediction studies |
| [16] | Penya, YK; Borges, CE; Agote, D.; Fernandez, I., Short-term load forecasting in airconditioned non-residential buildings, IEEE Int Symp Ind Electron, 2011, 1359-1364 (2011) |
| [17] | Chou JS, Bui DK (2014) : Modeling heating and cooling loads by artificial intelligence for energy-efficient building design. Energy Build 2014; 82:437-46 |
| [18] | Wang, R.; Lu, S.; Feng, W., A novel improved model for building energy consumption prediction based on model integration, Appl Energy, 262, 114561 (2020) · doi:10.1016/j.apenergy.2020.114561 |
| [19] | Chicco G, Di Somma M, Graditi G (2021) Overview of distributed energy resources in the context of local integrated energy systems. Distributed Energy Resources in local Integrated Energy Systems. Elsevier, pp 1-29 |
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. In some cases that data have been complemented/enhanced by data from zbMATH Open. This attempts to reflect the references listed in the original paper as accurately as possible without claiming completeness or a perfect matching.
