TU Darmstadt / ULB / TUbiblio

Machine Learning based System Identification Tool for data-based Energy and Resource Modeling and Simulation

Weber, Thomas and Sossenheimer, Johannes and Schäfer, Steffen and Ott, Moritz and Walther, Jessica and Abele, Eberhard (2019):
Machine Learning based System Identification Tool for data-based Energy and Resource Modeling and Simulation.
In: Procedia CIRP, 26th CIRP Conference on Life Cycle Engineering (LCE) Purdue University, West Lafayette, IN (USA), Elsevier B.V., pp. 683-688, 80, DOI: 10.1016/j.procir.2018.12.021,
[Article]

Abstract

Generated machine data is often not fully utilized in modern power production, although it could provide new approaches to significantly increase productivity, flexibility and resource efficiency as well as energy efficiency in production. Data-based models, which can be created with the help of machine learning algorithms, can map the system's behavior accurately and thus provide a basis for a better system understanding for further energy und resource optimization approaches. The objective of this paper is to develop a generic system identification tool that uses the above-mentioned data-based modeling approach to optimize the electrical power and resource consumption for a given load, regardless of the considered plant or machine. Therefore, the system identification tool autonomously preprocesses the data, compares different hyperparameters for neural networks to reproduce the system's behavior and finally selects the best-suited regression algorithm with the corresponding hyperparameters.

Item Type: Article
Erschienen: 2019
Creators: Weber, Thomas and Sossenheimer, Johannes and Schäfer, Steffen and Ott, Moritz and Walther, Jessica and Abele, Eberhard
Title: Machine Learning based System Identification Tool for data-based Energy and Resource Modeling and Simulation
Language: English
Abstract:

Generated machine data is often not fully utilized in modern power production, although it could provide new approaches to significantly increase productivity, flexibility and resource efficiency as well as energy efficiency in production. Data-based models, which can be created with the help of machine learning algorithms, can map the system's behavior accurately and thus provide a basis for a better system understanding for further energy und resource optimization approaches. The objective of this paper is to develop a generic system identification tool that uses the above-mentioned data-based modeling approach to optimize the electrical power and resource consumption for a given load, regardless of the considered plant or machine. Therefore, the system identification tool autonomously preprocesses the data, compares different hyperparameters for neural networks to reproduce the system's behavior and finally selects the best-suited regression algorithm with the corresponding hyperparameters.

Journal or Publication Title: Procedia CIRP, 26th CIRP Conference on Life Cycle Engineering (LCE) Purdue University, West Lafayette, IN (USA)
Volume: 80
Publisher: Elsevier B.V.
Uncontrolled Keywords: Energy efficiency; Ressource efficiency; System Identification; Machine Learning
Divisions: 16 Department of Mechanical Engineering
16 Department of Mechanical Engineering > Institute of Production Management, Technology and Machine Tools (PTW)
16 Department of Mechanical Engineering > Institute of Production Management, Technology and Machine Tools (PTW) > ETA Energy Technologies and Applications in Production
Date Deposited: 27 Aug 2019 05:31
DOI: 10.1016/j.procir.2018.12.021
Export:
Suche nach Titel in: TUfind oder in Google

Optionen (nur für Redakteure)

View Item View Item