Gast ::
| Hauptseite > Publikationsdatenbank > Inverse Deep Learning Raytracing for heliostat surface prediction |
| Hauptseite > Publikationsdatenbank > Inverse Deep Learning Raytracing for heliostat surface prediction |
| Journal Article | FZJ-2026-01063 |
; ; ; ; ;
2025
Elsevier Science
Amsterdam [u.a.]
This record in other databases: 
Please use a persistent id in citations: doi:10.1016/j.solener.2025.113312 doi:10.34734/FZJ-2026-01063
Abstract: Concentrating Solar Power (CSP) plants play a crucial role in the global transition toward sustainable energy. A key factor in ensuring the safe and efficient operation of CSP plants is the distribution of concentrated flux density on the receiver. However, the non-ideal flux density generated by individual heliostats can undermine the safety and efficiency of the power plant. The flux density from each heliostat is influenced by its precise surface, which includes factors such as canting and mirror errors. Accurately measuring these surfaces for a large number of heliostats in operation is a formidable challenge. Consequently, control systems often rely on the assumption of ideal surface conditions, which compromises both safety and operational efficiency. In this study, we introduce inverse Deep Learning Raytracing (iDLR), an innovative method designed to predict heliostat surfaces based solely on target images obtained during heliostat calibration. Our simulation-based investigation reveals that the flux density distribution of a single heliostat contains sufficient information to enable deep learning models to accurately predict the underlying surface with deflectometry-like precision in most cases, achieving a median Mean Absolute Error of approximately 0.14 mm). When integrating the iDLR surface predictions into a ray-tracing environment to compute flux densities, our method achieves an accuracy of 92%, surpassing the performance of the ideal heliostat assumption by 25%. Additionally, we assess the limitations of this method, particularly in relation to surface prediction accuracy and resultant flux density predictions. Furthermore, we present an innovative and efficient heliostat surface model based on NURBS. This approach achieves a highly compact representation, requiring only 256 parameters to define the surface—a reduction of 99.97% in the amount of parameter and a 99.91% in memory usage. This efficient model enables resource-effective deep learning for heliostat surface predictions, positioning it as a promising state-of-the-art solution for heliostat surface parameterization. Our findings demonstrate that iDLR has significant potential to optimize CSP plant operations, enhancing overall efficiency and increasing the energy output of power plants.
; 
; 
; Clarivate Analytics Master Journal List ; Current Contents - Engineering, Computing and Technology ; Ebsco Academic Search ; Essential Science Indicators ; IF >= 5 ; JCR ; SCOPUS ; Science Citation Index Expanded ; Web of Science Core Collection
|
The record appears in these collections: |
PDF