elib
DLR-Header
DLR-Logo -> http://www.dlr.de
DLR Portal Home | Imprint | Privacy Policy | Contact | Deutsch
Fontsize: [-] Text [+]

Performance Increase of Solar Power Plants by applying Deep Learning Algorithms for Heliostatfield Calibration

Pargmann, Max and Maldonado Quinto, Daniel (2019) Performance Increase of Solar Power Plants by applying Deep Learning Algorithms for Heliostatfield Calibration. Solar Paces 2020, 01.-04. Okt. 2019, Süd Korea.

[img] PDF
440kB

Abstract

The efficiency of a power plant is mainly influenced by the ability of its heliostats to track solar radiation biaxially to the receiver. In order to minimize the tracking error of heliostats, especially in large plants, it is essential to re-calibrate the heliostat control regularly for maximum power generation. Conventional calibration methods with regression can meet the requirements of frequent and regular use, but have large errors due to nonlinear behavior in heliostat mechanics and external factors such as wind or dust, so they must be performed relatively frequently. We present an improved regression method based on AI algorithms, by showing different ways of implementing these Algorithms into existing procedures without exchanging the hardware. and demonstrate that it is possible to reduce tracking errors, compared to common Algorithms, even with a small amount of data. Implementation of a KI controlled digital twin for steady improvement of solar power plant models Due to climate change, increasingly scarce fossil resources and the preservation of energy sovereignty, renewable energy is becoming more and more the focus of public attention. Solar tower power plants are key of the energy revolution, as they can provide a short-term storage function for heat and thus for more regular electricity production. Due to their nature, they also require more planning effort than e.g. photovoltaic systems. Numerical models are used not only in the layout of fields, but also in the control and regulation. These numerical models can represent reality, limited by the quantity of influencing factors, only within a certain model error. In order to reduce the influencing factors new sensors can be attached. However, this can be uneconomical as well as slowing down conventional machine-learning algorithms and thus rendering them unusable. we want to reduce the model and running time of the numerical models by using an AI controlled digital twin. For this purpose we train neural networks with real data collected during power plant operation. As a proof-of-concept We present an improved regression method to calibrate Heliostats based on AI algorithms, by showing different ways of implementing these Algorithms into existing procedures without exchanging the hardware. and demonstrate that it is possible to reduce tracking errors, compared to common Algorithms, even with a small amount of data.

Item URL in elib:https://elib.dlr.de/133539/
Document Type:Conference or Workshop Item (Poster)
Title:Performance Increase of Solar Power Plants by applying Deep Learning Algorithms for Heliostatfield Calibration
Authors:
AuthorsInstitution or Email of AuthorsAuthor's ORCID iD
Pargmann, MaxMax.Pargmann (at) dlr.deUNSPECIFIED
Maldonado Quinto, DanielDaniel.MaldonadoQuinto (at) dlr.deUNSPECIFIED
Date:October 2019
Refereed publication:Yes
Open Access:Yes
Gold Open Access:No
In SCOPUS:No
In ISI Web of Science:No
Status:Published
Keywords:Neural Networks, Solar Tower Power Plant, Deep Learning, Calibration, Heliostatfield
Event Title:Solar Paces 2020
Event Location:Süd Korea
Event Type:international Conference
Event Dates:01.-04. Okt. 2019
HGF - Research field:Aeronautics, Space and Transport
HGF - Program:Transport
HGF - Program Themes:Transport System
DLR - Research area:Transport
DLR - Program:V VS - Verkehrssystem
DLR - Research theme (Project):V - Energie und Verkehr
Location: Köln-Porz
Institutes and Institutions:Institute of Solar Research > Solar Power Plant Technology
Deposited By: Pargmann, Max
Deposited On:23 Jan 2020 10:46
Last Modified:05 Mar 2020 13:07

Repository Staff Only: item control page

Browse
Search
Help & Contact
Information
electronic library is running on EPrints 3.3.12
Copyright © 2008-2017 German Aerospace Center (DLR). All rights reserved.