Urban building energy retrofit (UBER) is a critical strategy for advancing the low-carbon and climate-resilience transformation of cities. The integration of machine learning (ML), data-driven clustering, and multi-objective optimization (MOO) is a key aspect of artificial intelligence (AI) that is transforming the process of retrofit decision-making. This integration enables the development of scalable, cost-effective, and robust solutions on an urban scale. This systematic review synthesizes recent advances in AI-driven MOO frameworks for UBER, focusing on how state-of-the-art methods can help to identify and prioritize retrofit targets, balance energy, cost, and environmental objectives, and develop transparent, stakeholder-oriented decision-making processes. Key advances highlighted in this review include the following: (1) the application of ML-based surrogate models for efficient evaluation of retrofit design alternatives; (2) data-driven clustering and classification to identify high-impact interventions across complex urban fabrics; (3) MOO algorithms that support trade-off analysis under real-world constraints; and (4) the emerging integration of explainable AI (XAI) for enhanced transparency and stakeholder engagement in retrofit planning. Representative case studies demonstrate the practical impact of these approaches in optimizing envelope upgrades, active system retrofits, and prioritization schemes. Notwithstanding these advancements, considerable challenges persist, encompassing data heterogeneity, the transferability of models across disparate urban contexts, fragmented digital toolchains, and the paucity of real-world validation of AI-based solutions. The subsequent discussion encompasses prospective research directions, with particular emphasis on the potential of deep learning (DL), spatiotemporal forecasting, generative models, and digital twins to further advance scalable and adaptive urban retrofit.

In recent years, there has been increasing concern about the effects of indoor thermal environments on human physical and mental health. This paper aimed to study the current status of the thermal environment and thermal comfort in the classrooms of Northeastern University during the heating season. The indoor thermal environment was analyzed with the use of field measurements, a subjective questionnaire, regression statistics, and the entropy weight method. The results show that personnel population density is an important factor affecting the temperature and relative humidity variations in classrooms. The results also show that the temperature and relative humidity in a lecture state are respectively 4.2 °C and 11.4% higher than those in an idle state. In addition, in university classrooms in Shenyang, the actual thermal neutral temperature is 2.5 °C lower than the predicted value of the Predicted Mean Vote. It was found that increasing indoor relative humidity can effectively improve the overall thermal comfort of subjects. Furthermore, the temperature preference of women was higher than that of men. Therefore, when setting the initial heating temperature, the personnel population density and sufficient indoor relative humidity have been identified as the key factors for improving the thermal environment of the classroom.

Shopping malls are large buildings and thus have extremely high heating or cooling costs and energy requirements. This study explored the response patterns of human physiological parameters to sudden temperature changes (temperature difference >30 °C) at moderate activity levels in a Shenyang shopping mall. The temperature was set to −12 °C for cold conditions and indoor temperature conditions were set to 18 and 24 °C. Fifty participants underwent hot—cold—hot room temperature exposure. The following results were observed for short-duration stays in the shopping mall. (1) When the temperature difference between hot and cold environments was >30 °C and the indoor temperature did not exceed 24 °C, 12–18 min on average was required for the body to reach a new physiological equilibrium. Children required less time to return to a steady state than other age-groups. (2) Children, adolescents, and young adults preferred indoor temperature conditions of 18 °C, whereas middle-aged adults preferred a slightly warmer environment. Thus, in view of the excessively high indoor temperature of the mall, its temperature should be maintained within a range that not only conserves energy but also provides human comfort.

In recent years, deep learning methods have been used with increasing frequency to solve architectural design problems. This paper aims to study the spatial functional layout of deep learning-assisted generation subway stations. Using the PointNet++ model, the subway station point cloud data are trained and then collected and processed by the author. After training and verification, the following conclusions are obtained: (1) the feasibility of spatial deep learning for construction based on PointNet++ in the form of point cloud data is verified; (2) the effectiveness of PointNet++ for the semantic segmentation and prediction of metro station point cloud information is verified; and (3) the results show that the overall 9:1 training prediction data have 60% + MIOU and 75% + accuracy for 9:1 training prediction data in the space of 20 × 20 × 20 and a block_size of 10.0. This paper combines the deep learning of 3D point cloud data with architectural design, breaking through the original status quo of two-dimensional images as research objects. From the dataset level, the limitation that research objects such as 2D images cannot accurately describe 3D space is avoided, and more intuitive and diverse design aids are provided for architects.

Achieving optimal daylighting is a cornerstone of sustainable architectural design, impacting energy efficiency and occupant well-being. Fast and accurate prediction during the conceptual phase is crucial but challenging. While physics-based simulations are accurate but slow, existing machine learning methods often rely on restrictive parametric inputs, limiting their application across free-form designs. This study presents a novel, geometry-agnostic framework that uses only building facade unfolding diagrams as input to a Generative Adversarial Network (GAN). Our core innovation is a 2D representation that preserves 3D facade geometry and orientation by “unfolding” it onto the floor plan, eliminating the need for predefined parameters or intermediate features during prediction. A Pix2pixHD model was trained, validated, and tested on a total of 720 paired diagram-simulation images (split 80:10:10). The model achieves high-fidelity visual predictions, with a mean Structural Similarity Index (SSIM) of 0.93 against RADIANCE/Daysim benchmarks. When accounting for the practical time of diagram drafting, the complete workflow offers a speedup of approximately 1.5 to 52 times compared to conventional simulation. This work provides architects with an intuitive, low-threshold tool for rapid daylight performance feedback in early-stage design exploration.

This paper aims to study the generation of spatial function layout of subway stations assisted by deep learning, and train the point cloud data of subway stations based on the Pointnet + + model in deep learning. The point cloud data of subway stations comes from the subway station data of large and medium-sized cities collected and processed by the author. After training and verification, the following conclusions are drawn: (1) It has been verified that the spatial deep learning of buildings in the form of point clouds is highly feasible, and the point cloud format of the architectural space model is fully compatible under the Pointnet + + model. (2) Verified the effectiveness of Pointnet + + for semantic segmentation and prediction of subway station cloud information. The results show that the predicted data has 60%+MIOU (MeanIoU average intersection and union ratio) and 75%+Acc (Accuracy). This paper uses an interdisciplinary research method to combine deep learning of 3D point cloud data with architectural design, breaking through the current situation of using 2D images as research objects, and avoiding the application of "deep learning" to 2D images. Objects cannot accurately describe the limitations of 3D space, providing architects with more intuitive and diverse design assistance.