There is a newer version of the record available.

Published July 7, 2026 | Version 1.0

Neural network models for image-to-image flux map corrections

Authors/Creators

  • 1. ROR icon Deutsches Zentrum für Luft- und Raumfahrt e. V. (DLR)

Description

In this publication, the models utilized for the operational flux density prediction using ray-tracing simulation as input are provided. Both the models and auxiliary functions (for inference or training) are included here. 

The Transformer models utilized for ray direction predictions - including the dimensional downsampling - are also included.

Files

Unet+AG.ipynb

Files (85.4 MB)

Name Size Download all
md5:c05b429003ea939c72f845f70f02d36a
11.7 kB Download
md5:37fbc7a47392e26ab349417b86d52818
41.6 kB Download
md5:b2cd800117cf6aa86627a44e73a07438
9.9 kB Download
md5:cfb3b44aabd537666bc62d3a6bacdb06
60.3 kB Download
md5:987d663c3aafe2bae862301b2e9d21eb
50.6 kB Download
md5:d48d20669572298b506b1fa0fff59d12
25.1 kB Download
md5:2bea8ca244e2afad65094b938ce1071a
20.6 kB Preview Download
md5:98963fcc04d5d0b86430755ce34d8133
9.4 MB Preview Download
md5:2a614dde546b18468d86a31935b3b210
1.4 kB Download
md5:7b986b023e66049453127a9e7b4996fd
9.4 MB Preview Download
md5:c63e2aea820f2e1a86651854db650438
9.8 MB Preview Download
md5:5f7a6fe617b1ef2d77965e6f7d846664
9.4 MB Preview Download
md5:a79ad9ef6853a2f281cbe7b1558fb239
9.5 MB Preview Download
md5:0f0325f2837e64c30991f4459c78b41b
9.4 MB Preview Download
md5:b8f7635868e3f5050ef9b686cfbb4f92
9.4 MB Preview Download
md5:fdee112e297cfad8b219e6c3228044e5
14.1 kB Download
md5:7c65963ec5456b16f0f7caca25f1e82f
9.4 MB Preview Download
md5:2b4aedbf062d3642d3dfd9aa29eea4a8
9.4 MB Preview Download
md5:e29b1c127ef3cde721d429ab5b9e8eb9
1.4 kB Download

Additional details

Related works

Is documented by
Journal article: 10.1016/j.rineng.2025.107096 (DOI)
Journal article: 10.1016/j.egyai.2026.100763 (DOI)
Requires
Dataset: 10.5281/zenodo.15173386 (DOI)
Dataset: 10.5281/zenodo.21235808 (DOI)
Dataset: 10.5281/zenodo.15740381 (DOI)

Funding

European Commission
TOPCSP - Towards Competitive, Reliable, Safe and Sustainable Concentrated Solar Power (CSP) Plants 101072537

Dates

Submitted
2026-07-07

Software

Programming language
Python
Development Status
Active