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

Andreas Lenk (), Marcus Vogt and Christoph Herrmann
Additional contact information
Andreas Lenk: Volkswagen AG, 38440 Wolfsburg, Germany
Marcus Vogt: Volkswagen AG, 38440 Wolfsburg, Germany
Christoph Herrmann: Chair of Sustainable Production & Life Cycle Engineering, Institute of Machine Tools and Production Engineering, Technical University of Braunschweig, 38106 Braunschweig, Germany

Energies, 2024, vol. 18, issue 1, 1-34

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

Keywords: energy; automotive production; prediction; machine learning (search for similar items in EconPapers)
JEL-codes: Q Q0 Q4 Q40 Q41 Q42 Q43 Q47 Q48 Q49 (search for similar items in EconPapers)
Date: 2024
References: View references in EconPapers View complete reference list from CitEc
Citations:

Downloads: (external link)
https://www.mdpi.com/1996-1073/18/1/2/pdf (application/pdf)
https://www.mdpi.com/1996-1073/18/1/2/ (text/html)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:gam:jeners:v:18:y:2024:i:1:p:2-:d:1551107

Access Statistics for this article

Energies is currently edited by Ms. Agatha Cao

More articles in Energies from MDPI
Bibliographic data for series maintained by MDPI Indexing Manager ().