Diffuse Horizontal Illuminance Estimation and Predictive Uncertainty Quantification for Daylighting Applications

Daylighting is a key strategy for reducing artificial lighting demand in buildings, but its effective use requires reliable information on the daylight resource. While radiometric solar variables are commonly monitored, photometric quantities such as diffuse horizontal illuminance (LDFH) are less frequently measured, despite their relevance for daylight-responsive lighting control, visual comfort, and energy-efficient building operation. Moreover, uncertainty is often not quantified, as conventional models generally provide only deterministic outputs. This study proposes an artificial-intelligence framework for estimating LDFH from diffuse horizontal irradiance, solar geometry, and zenith luminance, while also quantifying its uncertainty. A Type-1 Sugeno adaptive neuro-fuzzy inference system (ANFIS) is used for LDFH estimation, and an Interval Type-2 Sugeno extension provides sample-specific predictive intervals. A simplified model based only on DHI and sin(α) is also evaluated to provide an alternative for stations where zenith luminance measurements are unavailable. The approach was trained using 2024 measurements and externally validated on independent years from 2019 to 2023. The Type-1 Sugeno ANFIS achieved an nRMSE of 8.55%, an R² of 0.98, and an nMBE of 0.00% in training. During the independent test years, nRMSE ranged from 8.37% to 10.59%, R² from 0.97 to 0.98, and nMBE from -3.48% to 2.10%. For a target coverage probability of 95.45%, the IT2-Sugeno model achieved 95.10% coverage in training, with a mean expanded uncertainty of 4.06 klux. During testing, coverage ranged from 93.00% to 96.24%, with mean uncertainty values between 3.59 and 4.14 klux. The proposed approach therefore enables direct estimation of LDFH together with its associated uncertainty, supporting the assessment of diffuse daylight availability.

Authors

Institutions

Publication Details

Journal
Journal of Daylighting
Published
2026-11-13
DOI
https://doi.org/10.15627/jd.2026.40
Primary Topic
Solar Radiation and Photovoltaics
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Diffuse Horizontal Illuminance Estimation and Predictive Uncertainty Quantification for Daylighting Applications

María Isabel Dieste-Velasco
Journal of Daylighting
Solar Radiation and Photovoltaics
article

Diffuse Horizontal Illuminance Estimation and Predictive Uncertainty Quantification for Daylighting Applications

María Isabel Dieste-Velasco
article en

Abstract

Daylighting is a key strategy for reducing artificial lighting demand in buildings, but its effective use requires reliable information on the daylight resource. While radiometric solar variables are commonly monitored, photometric quantities such as diffuse horizontal illuminance (LDFH) are less frequently measured, despite their relevance for daylight-responsive lighting control, visual comfort, and energy-efficient building operation. Moreover, uncertainty is often not quantified, as conventional models generally provide only deterministic outputs. This study proposes an artificial-intelligence framework for estimating LDFH from diffuse horizontal irradiance, solar geometry, and zenith luminance, while also quantifying its uncertainty. A Type-1 Sugeno adaptive neuro-fuzzy inference system (ANFIS) is used for LDFH estimation, and an Interval Type-2 Sugeno extension provides sample-specific predictive intervals. A simplified model based only on DHI and sin(α) is also evaluated to provide an alternative for stations where zenith luminance measurements are unavailable. The approach was trained using 2024 measurements and externally validated on independent years from 2019 to 2023. The Type-1 Sugeno ANFIS achieved an nRMSE of 8.55%, an R² of 0.98, and an nMBE of 0.00% in training. During the independent test years, nRMSE ranged from 8.37% to 10.59%, R² from 0.97 to 0.98, and nMBE from -3.48% to 2.10%. For a target coverage probability of 95.45%, the IT2-Sugeno model achieved 95.10% coverage in training, with a mean expanded uncertainty of 4.06 klux. During testing, coverage ranged from 93.00% to 96.24%, with mean uncertainty values between 3.59 and 4.14 klux. The proposed approach therefore enables direct estimation of LDFH together with its associated uncertainty, supporting the assessment of diffuse daylight availability.

Journal of DaylightingVol. 13(2)
Universidad de Burgos (ES)
Solar Radiation and Photovoltaics
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon