An Approach to Predicting Energy Demand Within Automobile Production Using the Temporal Fusion Transformer Model DOI Creative Commons

Andreas Lenk,

Marcus Vogt, Christoph Herrmann

et al.

Energies, Journal Year: 2024, Volume and Issue: 18(1), P. 2 - 2

Published: Dec. 24, 2024

The increasing share of renewable energies within energy systems leads to an increase in complexity. growing complexity is due the diversity technologies, ongoing technological innovations, and fluctuating electricity production. To continue ensure a secure, economical, needs-based supply, additional information needed efficiently control these systems. This impacts public industrial supply systems, such as vehicle factories. paper examines influencing factors applicability Temporal Fusion Transformer (TFT) model for weekly demand forecast at automobile production site. Seven different TFT models were trained demand. Six predicted electricity, heat, natural gas. Three used rolling day-ahead forecast, three entire week one step. In seventh model, was again, with target values being same model. analysis shows that forecasting method MAPE 13% already delivers good results predicting electrical prediction accuracy achieved sufficient use outcomes basis operational planning reporting. However, further improvements are still required automated system reduce procurement costs. heat gas demands show too high deviations, 62% 39% accurately predict demands, must be identified explain

Language: Английский

A state-of-the-art comparative review of load forecasting methods: Characteristics, perspectives, and applications DOI Creative Commons

Mahmudul Hasan,

Zannatul Mifta,

Sumaiya Janefar Papiya

et al.

Energy Conversion and Management X, Journal Year: 2025, Volume and Issue: unknown, P. 100922 - 100922

Published: Feb. 1, 2025

Language: Английский

Citations

0
AutoPQ: Automating quantile estimation from point forecasts in the context of sustainability DOI Creative Commons
Stefan Meisenbacher, Kaleb Phipps, Oskar Taubert

et al.

Applied Energy, Journal Year: 2025, Volume and Issue: 392, P. 125931 - 125931

Published: April 23, 2025

Language: Английский

Citations

0
Transformer training strategies for forecasting multiple load time series DOI Creative Commons
Matthias Hertel, Maximilian Beichter, Benedikt Heidrich

et al.

Energy Informatics, Journal Year: 2023, Volume and Issue: 6(S1)

Published: Oct. 19, 2023

Abstract In the smart grid of future, accurate load forecasts on level individual clients can help to balance supply and demand locally prevent outages. While number monitored will increase with ongoing meter rollout, amount data per client always be limited. We evaluate whether a Transformer forecasting model benefits from transfer learning strategy, where global univariate is trained time series multiple clients. experiments two datasets containing several hundred clients, we find that training strategy superior multivariate local strategies used in related work. On average, results 21.8% 12.8% lower errors than other strategies, measured across horizons one day month into future. A comparison linear models, multi-layer perceptrons LSTMs shows Transformers are effective for when they strategy.

Language: Английский

Citations

10
Using conditional Invertible Neural Networks to perform mid‐term peak load forecasting DOI Creative Commons
Benedikt Heidrich, Matthias Hertel, Oliver Neumann

et al.

IET Smart Grid, Journal Year: 2024, Volume and Issue: 7(4), P. 460 - 472

Published: April 26, 2024

Abstract Measures for balancing the electrical grid, such as peak shaving, require accurate forecasts lower aggregation levels of loads. Thus, Big Data Energy Analytics Laboratory (BigDEAL) challenge—organised by BigDEAL—focused on forecasting three different daily characteristics in low aggregated load time series. In particular, participants challenge were asked to provide long‐term with horizons up 1 year qualification. The authors present approach KIT‐IAI team from Institute Automation and Applied Informatics at Karlsruhe Technology. is based a hybrid generative model. use conditional Invertible Neural Network (cINN). cINN gets forecast sliding mean representative trend, weather features, calendar information conditioning input. By this, proposed method achieved second place overall won two out tracks BigDEAL challenge.

Language: Английский

Citations

3
Kolmogorov–Arnold recurrent network for short term load forecasting across diverse consumers DOI
Muhammad Umair Danish, Katarina Grolinger

Energy Reports, Journal Year: 2024, Volume and Issue: 13, P. 713 - 727

Published: Dec. 24, 2024

Language: Английский

Citations

3
Self-attention mechanism to enhance the generalizability of data-driven time-series prediction: A case study of intra-hour power forecasting of urban distributed photovoltaic systems DOI
Hanxin Yu,

Shanlin Chen,

Yinghao Chu

et al.

Applied Energy, Journal Year: 2024, Volume and Issue: 374, P. 124007 - 124007

Published: July 31, 2024

Language: Английский

Citations

2
Feasibility of Forecasting Highly Resolved Power Grid Frequency Utilizing Temporal Fusion Transformers DOI
Sebastian Pütz, Hadeer El Ashhab, Matthias Hertel

et al.

Published: May 31, 2024

Language: Английский

Citations

1
Digital Twins: Case Study of Energy Metaverse and Edge-Cloud Integration DOI
Ali Aghazadeh Ardebili, Angelo Martella, Cristian Martella

et al.

2021 IEEE International Conference on Big Data (Big Data), Journal Year: 2024, Volume and Issue: unknown, P. 5476 - 5485

Published: Dec. 15, 2024

Language: Английский

Citations

1
Transfer Learning in Transformer-Based Demand Forecasting For Home Energy Management System DOI
Gargya Gokhale, Jonas Van Gompel, Bert Claessens

et al.

Published: Nov. 3, 2023

Increasingly, homeowners opt for photovoltaic (PV) systems and/or battery storage to minimize their energy bills and maximize renewable usage. This has spurred the development of advanced control algorithms that maximally achieve those goals. However, a common challenge faced while developing such controllers is unavailability accurate forecasts household power consumption, especially shorter time resolutions (15 minutes) in data-efficient manner. In this paper, we analyze how transfer learning can help by exploiting data from multiple households improve single house's load forecasting. Specifically, train an forecasting model (a temporal fusion transformer) using different households, then finetune global on new with limited (i.e., only few days). The obtained models are used consumption next 24 hours (day-ahead) at resolution 15 minutes, intention these as Model Predictive Control. We show benefit setup versus solely individual household's data, both terms real-world data.

Language: Английский

Citations

3
Transfer Learning for Day-Ahead Load Forecasting: A Case Study on European National Electricity Demand Time Series DOI Creative Commons
Alexandros Menelaos Tzortzis, Sotiris Pelekis, Evangelos Spiliotis

et al.

Mathematics, Journal Year: 2023, Volume and Issue: 12(1), P. 19 - 19

Published: Dec. 21, 2023

Short-term load forecasting (STLF) is crucial for the daily operation of power grids. However, non-linearity, non-stationarity, and randomness characterizing electricity demand time series renders STLF a challenging task. Various approaches have been proposed improving STLF, including neural network (NN) models which are trained using data from multiple that may not necessarily include target series. In present study, we investigate performance special case namely transfer learning (TL), by considering set 27 represent national day-ahead indicative European countries. We employ popular easy-to-implement feed-forward NN model perform clustering analysis to identify similar patterns among enhance TL. this context, two different TL approaches, with without step, compiled compared against each other as well typical training setup. Our results demonstrate can outperform conventional approach, especially when techniques considered.

Language: Английский

Citations

3