Tabatabaei Manesh, M., Rajaian Hoonejani, M., Ghafari Gousheh, S., Abdolmaleki, A., Nikkhah Dehnavi, A., & Shahrashoob, A. (2025). AI-driven control algorithm using machine learning and genetic optimization for enhancing visual comfort in adaptive façades. Automation in Construction, 179, Article 106474. https://doi.org/10.1016/j.autcon.2025.106474.
View Publication
Abstract
Effective management of daylight and visual comfort in office spaces remains a challenge, as existing shading systems often lack adaptability to changing environmental conditions and occupant needs. This paper presents an AI-driven real-time shading control algorithm that optimizes visual comfort using machine learning-based surrogate models and evolutionary optimization. A non-conventional adaptive façade was simulated using Radiance and Ladybug Tools across nine U.S. climates. Four machine learning models were evaluated for predicting Task Illuminance (Et) and Vertical Eye Illuminance (Ev), with Extra Trees achieving the highest accuracy (R2
= 0.95). A Non-dominated Sorting Genetic Algorithm II (NSGA-II) balances glare reduction and daylight utilization by optimizing façade configurations in real time. In contrast to prior approaches constrained to fixed geometries and single-objective control, this paper introduces a generalizable multi-objective control framework. Results show that AI-driven optimization significantly improves adaptive façade performance, offering a scalable solution for intelligent daylight and comfort management.
Keywords
Smart façade control; Machine learning; Surrogate models; Visual comfort; Task illuminance; Vertical eye illuminance; Dynamic shading
Rashtian, Z., Manesh, M. T., Tahsildoost, M., & Zomorodian, Z. S. (2025). Data-driven real-time visualization of urban heat islands using mean radiant temperature for urban design. Energy and Buildings, 115470-. https://doi.org/10.1016/j.enbuild.2025.115470
View Publication
Abstract
The Mean Radiant Temperature (Tmrt), is a critical indicator for understanding urban thermal comfort and microclimate conditions, particularly in urban areas experiencing higher temperatures compared to rural surroundings. Tmrt is heavily influenced by urban morphology, including building layout, street design, and green spaces which alter airflow, shading, and heat retention. Evaluating geometry alternatives during the early design stages in urban neighborhoods is challenging due to lengthy simulations and the need for extensive expertise in physical models. Recent studies have employed data-driven methods for quick design comparisons and new urban layout evaluations, successfully predicting Thermal indicators of Urban heat Island phenomenon but often limited by the diversity of urban configurations inputs used in training datasets. To address these limitations, this study proposes a novel framework that uses machine learning models to predict Tmrt as the primary indicator. A comprehensive training dataset of 200 cases was generated in Rhino7 using Grasshopper, Ladybug, and Dragonfly plugins. Sensitivity analysis was conducted to assess the impact of input uncertainties on model predictions, and the model’s performance was validated against unseen configurations. Among six machine learning algorithms tested, the CatBoost Regressor achieved the highest accuracy, predicting Tmrt with an R2 = 0.93, RMSE = 4.30 °C, and MAE = 2.34 °C. Validation using 20 additional cases showed an accuracy of R2 = 0.71, RMSE = 3.34 °C, and MAE = 2.27 °C in predicting Tmrt heat maps for new urban configurations. This framework successfully enables real-time Tmrt heat map analysis in simplified cubic neighborhoods within a 3D environment. Additionally, it enhances the temporal and spatial resolution of thermal patterns predictions, offering rapid and detailed insights into various urban design alternatives.
Tabatabaei Manesh, M., Nikkhah Dehnavi, A., Tahsildoost, M., & Alambeigi, P. (2024). Acoustic design evaluation in educational buildings using artificial intelligence. Building and Environment, 261, 111695-. https://doi.org/10.1016/j.buildenv.2024.111695
View Publication
Abstract
Speech intelligibility is a critical aspect of building science, particularly in educational buildings where poor sound quality may have a detrimental impact on students' learning and teachers’ health. However, considering the numerous building regulations proposing varying definitions and ranges of acoustic comfort, calculating the necessary acoustic indicators can be challenging for designers. Speech intelligibility is a crucial component of indoor acoustics and acoustic comfort and can be calculated using formulas, simulation software, and data-based web tools. While formulas are fast, they lack details; acoustic simulation software is highly accurate but time-consuming and expensive. Data-based web tools, including machine learning algorithms, offer both speed and accuracy and are widely accessible. In this study, we present a system utilizing machine learning techniques to predict acoustic indicators, numeric and heatmap, in an educational building. The Pachyderm plugin in the Grasshopper was utilized to conduct extensive simulations in a single educational space, focusing on acoustic indicators in six different frequencies and general modes. Using Catboost and the pix2pix algorithm, the prediction models provide numerical and image indices on the developed dataset. Also, SHAP values were employed to interpret the Catboost model, analyzing the significance of each feature. The results showed remarkable accuracy, (i.e., between 89 % and 99 %) in the numerical portion, and PSNR index ranging from 0.817 to 0.970, and an SSIM index ranging from 15.56 to 31.57 in the image section. By utilizing data-driven methods, the system provides an efficient and accurate approach to calculating acoustic indicators, helping to ensure optimal acoustic environment in educational buildings.
Keywords
Building acoustics; Catboost; Pix2pix; Educational building; Speech intelligibility
Tabatabaei Manesh, M., Nikkhah Dehnavi, A., & Rajaian, M. (2024). Using Machine Learning To Predict And Visualize Acoustic Quality In Educational Buildings. The 2024 International ConCave Ph.D. Symposium: Divergence in Architectural Research. Georgia Tech, Atlanta, April 4-5.
View Publication
Mohammad Tabatabaei Manesh is a computational designer and building science researcher with expertise in programming and building performance. He works on the application of machine learning and deep learning in building performance, developing web apps and tools for architects. Currently, Mohammad’s work focuses on the design, fabrication, and evaluation of acoustic metamaterials for the built environment.