Wenjie Liao

The advancement of generative artificial intelligence (AI) has accelerated the progress of intelligent design methods for building structures. However, generative AI models struggle to design inclined shear wall components due to their limited generalization ability. To address this obstacle, this study proposes a data augmentation method for the planar layout of rotated shear wall structures. By enhancing the dataset and associated post-processing methods, the generalization ability of design methods driven by generative AI algorithms is effectively improved. Specifically, an augmented dataset by rotating non-inclined component design data was constructed, with rotation angles from 0° to 60°, to encompass inclined shear wall design scenarios. The improved generalization performance between diffusion models, generative adversarial networks, and graph neural networks was then compared. An automatic vectorization extraction method for inclined components from the generated design feature tensors is established, enabling the application of the enhanced generative AI algorithms. Moreover, algorithm performance analysis and typical case studies show that the diffusion model performs best in the inclined shear wall design task. This comprehensive analysis confirms that data augmentation significantly improves the adaptability of generative AI to inclined component design, providing a valuable reference for enhancing the generalization of data-driven AI design in building structures.

Rapid evaluation of regional earthquake-induced landslides (EQILs) plays a pivotal role in the efficiency of emergency responses following seismic events. Diverging from the traditional EQIL prediction approaches that rely on ground-motion intensity measures (GMIMs), this research introduces an open-source rapid regional EQIL evaluation technique utilizing time-history analysis (THA). This open-source framework integrates the ShakeMap methodology to construct the ground-motion response spectrum for the designated area, followed by the use of a continuous wavelet transform to adjust ground motions from nearby stations and establish the ground-motion field. Subsequently, the THA-based Newmark rigid sliding block (NRSB) method is applied to estimate the landslide probability. This open-access approach ensures transparency, reproducibility, and scalability for practical applications. To demonstrate the utility of this proposed technique, the 2004 Niigata Chuetsu earthquake serves as a case study. The key findings are summarized as follows: (1) The ShakeMap-based approach to generating the ground-motion field effectively provides the seismic input needed for EQIL evaluation. (2) The THA-based evaluation yields results that more closely align with the observed earthquake impacts, in comparison to methods dependent on GMIMs. (3) The proposed open-source method facilitates the acquisition of EQIL evaluation outcomes shortly after an earthquake, offering a valuable and accessible tool for the expeditious assessment of EQILs.

Generative artificial intelligence (AI) applications in structural design face persistent challenges due to training data limitations, particularly datasets that lack compliance with critical physical and material requirements. This study proposes a structural optimization-based data enhancement method to address quality deficiencies in generative AI training data, specifically targeting shear wall layout design through diffusion-based generative models. The proposed method introduces three key innovations: (1) A novel generative AI design workflow incorporating data enhancement phase and modifying data preparation and model evaluation phases; (2) Regression formulas enabling data enhancement with incomplete design information through feature distribution analysis; (3) A shear wall layout optimization method for simultaneously improving physical and material metrics. Experimental validation reveals marked improvements in design outcomes through enhanced training data. Specifically, physically non-compliant structural designs show a 67% reduction in occurrence frequency, while material costs for compliant designs decrease by 0.5%. Additionally, performance consistency improves significantly. By addressing data quality limitations through structural optimization, this approach enhances the practical viability of diffusion models in structural engineering applications while preserving adaptability to advanced AI algorithms. The proposed method is modular and scalable, offering potential for extension to other structural systems (e.g., steel frames, composite structures) and design challenges, thereby advancing AI-driven innovation in structural engineering.

As buildings and structures age, the challenges of reinforcement and retrofitting become more significant, especially as their service life extends and the demand for seismic fortification increases. Integrating buckling‐restrained braces (BRBs) is an effective retrofit technique; however, this approach requires multiple iterations of layout adjustments and mechanical performance analysis, which are highly dependent on engineers' design expertise, resulting in low efficiency. To address this, the study proposes a two‐stage intelligent retrofit design method that integrates generative Artificial intelligence (AI) techniques with optimization algorithms for reinforced concrete (RC) frame structures using BRBs: (1) a diffusion model‐based potential BRB layout generation stage, and (2) an online learning algorithm‐based design optimization stage. In Stage 1, a diffusion model was employed to analyze architectural characteristics, identify potential BRB locations, narrow the feasible solution space for the optimization process, and ensure that the design meets empirical constraints. In Stage 2, an optimization algorithm, integrated with mechanical performance evaluation, was employed to determine the optimal locations and sizes of BRBs. Case studies revealed that these two methods enhanced efficiency by approximately 50 times compared to the direct design by engineers while maintaining design rationality and safety. Overall, these results demonstrate the feasibility and generalizability of the method in practical engineering applications, offering a reference for the intelligent design of more complex structural retrofits in the future.

The generation and editing of floor plans are critical in architectural planning, requiring a high degree of flexibility and efficiency. Existing methods demand extensive input information and lack the capability for interactive adaptation to user modifications. This paper introduces ChatHouseDiffusion, which leverages large language models (LLMs) to interpret natural language input, employs graphormer to encode topological relationships, and uses diffusion models to flexibly generate and edit floor plans. This approach allows iterative design adjustments based on user ideas, significantly enhancing design efficiency. Compared to existing models, ChatHouseDiffusion achieves higher Intersection over Union (IoU) scores, permitting precise, localized adjustments without the need for complete redesigns, thus offering greater practicality. Experiments demonstrate that our model not only strictly adheres to user specifications but also facilitates a more intuitive design process through its interactive capabilities.

To address the issue of costly computational expenditure related to high‐fidelity numerical models, surrogate models have been widely used in various engineering tasks, including design optimization. Despite the successful application of the existing surrogate models, physics‐based models depend largely on simplifications and assumptions, which render parameter calibration challenging; whereas data‐driven models require substantial data to reach their full potential, with their performance often being constrained in tasks when obtaining massive data is difficult. In this study, a hybrid surrogate model is proposed combining physics‐based and data‐driven models to rapidly estimate building seismic responses. The application of this model is exemplified through effective estimation of inter‐story drift ratios (IDRs), being a critical factor in shear‐wall structure design. Initially, a data augmentation technique and a parametric modeling procedure are introduced to significantly enhance the dataset diversity. Subsequently, a task decomposition strategy is proposed to effectively integrate a data‐driven graph neural network (GNN) and a physics‐based flexural‐shear model. Additionally, the output layer and the loss function of the GNN are modified to enhance the estimation accuracy by eliminating fundamental errors. Results of numerical experiments indicate that the proposed hybrid model can complete IDR estimations in an average time of 0.56 s, with a mean absolute percentage error of 12.7%. This performance significantly surpasses that of existing purely data‐driven and physics‐based models. A case study shows that the efficiency of the proposed hybrid model is approximately 100 times greater than that of conventional finite element software. This enables an accurate assessment of the design compliance with code requirements. The results of this study can be applied to the design optimization of seismic‐resistant building structures.

The rapid advancement of intelligent design technology in building structures has been primarily implemented in engineering practice through the use of local or cloud-based software to offer intelligent design services. However, local intelligent design services are time-consuming and require high-end hardware, whereas cloud-based designs fail to integrate seamlessly with existing design processes. Consequently, providing convenient intelligent design support for engineering practices is challenging. To address these problems, this study proposes a local–cloud collaborative intelligent design technology called AIstructure-Copilot, which serves as a structural intelligent design assistant. In this system, the local end performs routine graphical operations that align with engineers' design habits, whereas the cloud end executes generative artificial intelligence (AI) for intelligent design, thereby enhancing efficiency and effectively combining the strengths of both services. Specifically, this technology achieves a high level of automation and intelligence throughout the entire process, encompassing architectural design, structural design, and the establishment and execution of structural analysis models. This is accomplished by constructing a local–cloud collaborative mode, introducing a comprehensive data transmission format, and developing a cloud interface for generative AI algorithms. The effectiveness of the AIstructure-Copilot model was validated using a typical case study. The results demonstrate that AI design improves design efficiency by more than tenfold, satisfies the regulatory requirements of design schemes, and exhibits a discrepancy of approximately 20% when compared with designs created by competent engineers.