Search

Deep learning–driven intelligent generative design for building structures provides novel insights into intelligent construction. In a structural scheme design, the cross-sectional design of the shear wall components is critical. However, the current manual method is time-consuming and labor-intensive, and a statistical regression–based design is insufficiently accurate. Satisfying the requirements of a complex shear wall design in the real world is difficult for both methods. Generative adversarial networks (GANs) can extract implicit design laws by learning from design data and conduct end-to-end design effectively and rapidly. Although GANs have been adopted for intelligent structural design, some design rules established by competent engineers are difficult to capture. Hence, this study developed and subsequently adopted a rule-embedded GAN called StructGAN-Rule to address the demand for a rapid and accurate cross-sectional design of shear wall components. Specifically, a representation method that integrates design images and multiple design conditions was first established, which was followed by the construction of the training data set. Subsequently, based on the design rules, a differentiable tensor operator was built as a rule evaluator, which was embedded in the GAN to guide and constrain the training process. Finally, following the training of StructGAN-Rule, intelligent generative cross-sectional design based on the developed postprocessing method was effectively completed. Case studies on typical shear wall structures demonstrated that the StructGAN-Rule design satisfied the rule constraints well and was highly consistent with the design of engineers (approximately 1% difference). Moreover, the design efficiency was improved 6–10 times compared with that of the latter. [ABSTRACT FROM AUTHOR]

Seismic isolation can significantly improve the seismic resilience of buildings, resulting in a growing demand for seismic isolation designs. Meanwhile, the deep generative network‐based intelligent design can significantly increase scheme design efficiency. However, the performance of existing intelligent scheme designs is constrained by data quality and quantity. The limited availability of isolation design data hinders the development of intelligent seismic isolation design. Therefore, there is an emerging demand to establish an intelligent scheme design method that is free from data constraints and that can learn the physical mechanism and design rules. Consequently, this study proposes a physics‐rule‐co‐guided self‐supervised generative adversarial network (GAN) that can generate the layout and parameters of seismic isolation bearings by inputting the layout drawings of the shear wall structures. The critical physics‐rule‐co‐guided network model consists of a physics estimator, rule evaluator, discriminator, and design generator. The physics estimator is a deep neural network‐based surrogate model for predicting the mechanical response of an isolated structure, whereas the rule evaluator is a tensor operation‐based loss calculator that considers design rules. Furthermore, the proposed GAN model masters the schematic design ability of the seismic isolation of shear wall structures through multiphase hybrid learning of the pseudo‐labels, physical mechanism, and isolation design rules, obviating the need for ground‐truth data. Case studies also prove the rationality of the method, where the design results can effectively meet the code requirements and reduce the seismic response of the structure. [ABSTRACT FROM AUTHOR]

The intelligent design method based on generative adversarial networks (GANs) represents an emerging structural design paradigm where design rules are not artificially defined but are directly learned from existing design data. GAN-based methods have exhibited promising potential compared to conventional methods in the schematic design phase of reinforced concrete (RC) shear wall structures. However, for the following reasons, it is challenging to apply GAN-based approaches in the industry and to integrate them into the structural design process. (1) The data form of GAN-based methods is heterogeneous from that of the widely used computer-aided design (CAD) methods, and (2) GAN-based methods have high requirements on the hardware and software environment of the user's computer. As a result, this study proposes an integrated schematic design method for RC shear wall structures, providing a workable GAN application strategy. Specifically, (1) a preprocessing method of architectural CAD drawings is proposed to connect the GAN with the upstream architectural design; (2) a user-friendly cloud design platform is built to reduce the requirements of the user's local computer environment; and (3) a heterogeneous data transformation method and a parametric modeling procedure are proposed to automatically establish a structural analysis model based on GAN's design, facilitating downstream detailed design tasks. The proposed method makes it possible for the entire schematic design phase of RC shear wall structures to be intelligent and automated. A case study reveals that the proposed method has a heterogeneous data transformation accuracy of 97.3% and is capable of generating shear wall layout designs similar to the designs of a competent engineer, with 225 times higher efficiency. [ABSTRACT FROM AUTHOR]

Intelligent structural design using generative adversarial networks (GANs) is a revolutionary design approach for building structures. Despite its far‐reaching capability, the data quantity and quality may have limited the performance of such a data‐driven network. This study proposes to enhance the objectiveness of training processes by innovatively introducing a surrogate model, Physics Estimator, that informs the generator by appraising the physical behavior of the generated design. Dual loss functions evaluated by a traditional data‐driven discriminator and the Physics Estimator collaboratively foster the physics‐enhanced GAN architecture. We further develop a structural mechanics model to train and optimize the inherent accuracy of the Physics Estimator. The comparative study suggests that the proposed physics‐enhanced GAN can generate structural designs from architectural drawings and specified design conditions 44% better than a data‐driven design method and 90 times faster than a competent engineer. [ABSTRACT FROM AUTHOR]

Beam placement in shear wall systems is crucial in transferring vertical loads from floors to shear walls, ensuring structural integrity and optimal performance. Existing solutions using deep generative algorithms rely on pixel images and involve many model parameters, resulting in high computational costs. To address this issue, this paper presents a method based on graph neural networks (GNNs) with robust topological feature extraction capabilities. The method generates potential beam layout scenarios by incorporating architectural layouts, devising scheme design inputs and leveraging engineering experience. Adopting the proposed approach reduces the number of beam layout scenarios and significantly improves computation efficiency. The efficacy is demonstrated through various test cases, suggesting that the beam layouts designed by the proposed method closely resemble those by engineers. • An approach for designing beam layouts in shear wall structures that can account for topological features is presented. • A graph representation method for beam layout scenarios of shear wall structures is proposed. • A method for generating potential beam layout scenarios is presented. [ABSTRACT FROM AUTHOR]

Structural scheme design of shear wall structures is important because it is the first stage that guides the project along its entire structural design process and significantly impacts the subsequent design stages. Design methods for shear wall layouts based on deep generative algorithms have been proposed and achieved some success. However, current generative algorithms rely on pixel images to design shear wall layouts, which have many model parameters and require intensive calculations. Moreover, it is challenging to use pixel image-based methods to reflect the topological characteristics of structures and connect them with the subsequent design stages. The above defects can be effectively solved by representing a shear wall structure in graph data form and adopting graph neural networks (GNNs), which have a robust topological-characteristic-extraction capability. However, there is no existing research using GNN methods in the design of shear wall structures owing to the lack of graph representation methods and high-quality structural graph data for shear walls. Therefore, this study develops an intelligent design method for shear wall layouts based on GNNs. Two graph representation methods for a shear wall structure—graph edge representation and graph node representation—are examined. A data augmentation method for shear wall structures in graph data form is established to enhance the universality of the GNN performance. An evaluation method for both graph representation methods is developed. Case studies show that the shear wall layout designed using the established GNN method is highly similar to the design by experienced engineers. [ABSTRACT FROM AUTHOR]

• An attention-enhanced generative adversarial network is proposed for the layout design of shear walls. • A pre-training method is introduced to overcome the limitation of data shortage. • The proposed method significantly improves the shear wall layout of critical zones. • The design results of the proposed method exhibit better structural performance and lesser material consumption than existing methods. The preliminary layout design of the shear wall is critical for the design of reinforced concrete shear wall structures. Deep learning-based methods can learn design experience from existing design data and generate new designs efficiently. However, existing research is insufficient for the design of the local layout of shear walls in critical zones. The attention mechanism can identify the critical zones in an image that must be focused on and provides a new potential solution for improving the local design of shear wall structures. Therefore, this study proposed an attention-enhanced generative adversarial network model, named StructGAN-AE, for the intelligent design of shear wall structures. Additionally, a pre-training method was proposed to overcome the limitation of data shortage. Case studies showed that, compared with existing models, StructGAN-AE could generate a more reasonable local layout of shear walls in critical zones, such as elevator shafts, and the rationality of the generated shear walls was improved. The comprehensive evaluation metric of shear wall design results was improved by up to 30%. In terms of structural performance, the shear wall structures designed using StructGAN-AE demonstrated better anti-torsion performance, closer to engineers' design, and lower consumption of concrete and steel than those designed using existing methods. [ABSTRACT FROM AUTHOR]

• Intelligent generative structural design method based on the text and images guidance. • Novel generative adversarial networks fusing text and image to synthesis image. • Few-shot learning for text and image features extraction, fusion, and generation. • Universally high performance also in other designs, additional to structural design. Like the way engineers designing buildings, competent generative design methods try to understand the prescriptive requirement in text and architectural sketches, apply engineering principles and develop the structural design. However, this requirement may be challenging to existing methods because they are not good at simultaneously taking text and image input and then generating designs. This study proposed an innovative design approach, TxtImg2Img, to overcome the difficulties. Based on generative adversarial networks architecture, the generator is proposed to encode, extract and fuse texts and images, and generate new design images; the discriminator is developed to judge real and fake images and texts. Consequently, TxtImg2Img is advantageous in extracting features from the multimodal text and image data, fusing the features using the Hadamard product, and generating designs to satisfy the text-image requirements after learning from a limited number of design samples. Specifically, TxtImg2Img can generate structural design images without distortion, and the corresponding structural design meets the mechanical requirements, after being trained by dozens of words and hundreds of image data. The case studies confirm performance improvement of up to 21% and that the proposed approach presents a promising breakthrough for intelligent construction. [ABSTRACT FROM AUTHOR]

Artificial intelligence is reshaping building design processes to be smarter and automated. Considering the increasingly wide application of shear wall systems in high-rise buildings and envisioning the massive benefit of automated structural design, this paper proposes a generative adversarial network (GAN)-based shear wall design method, which learns from existing shear wall design documents and then performs structural design intelligently and swiftly. To this end, structural design datasets were prepared via abstraction, semanticization, classification, and parameterization in terms of building height and seismic design category. The GAN model improved its shear wall design proficiency through adversarial training supported by data and hyper-parametric analytics. The performance of the trained GAN model was appraised against the metrics based on the confusion matrix and the intersection-over-union approach. Finally, case studies were conducted to evaluate the applicability, effectiveness, and appropriateness of the innovative GAN-based structural design method, indicating significant speed-up and comparable quality. • An generative adversarial network-based automated structural design framework. • An open-access datasets of structural design drawings. • The datasets pre-processing method via pioneering abstraction, semanticization, classification, and parameterization. • Model validation approach based on confusion matrix and intersection-over-union metrics. [ABSTRACT FROM AUTHOR]

The generative intelligent design of building structures has been a rapidly evolving field in recent years, with shear wall structures efficiently designed using generative adversarial networks (GANs). Nevertheless, using RGB images to represent structural design drawings may not accurately capture the correlations and distinctions between building components, leading to shear-wall structure design inaccuracies. To address these challenges, this study proposes an optimized data representation and understanding method for the intelligent design of shear wall structures. Specifically, a component-based feature space data representation method was introduced to achieve a more accurate description of drawing features. A feature mask was also employed to precisely locate the key target areas for generating shear walls, enhancing the model's understanding of the data features. Additionally, a GAN with global and local discriminators was established to enhance the neural network's ability to generate shear wall designs at both the global image and local component levels. Finally, a training method for a GAN with dual discriminators based on feature masks is proposed. Experimental analyses and case studies showed that, compared with existing widely used GAN-based designs, the structural designs generated by the GAN trained with the proposed data representation and understanding methods demonstrated superior consistency with engineer-designed structures and outperformed structural seismic resistance. • Optimized data representation and understanding method for structural design. • Improve the model's understanding of component relationships in drawings. • Enhance performance on data with sparse features. • Improve the model's ability for joint learning of both global and local design. [ABSTRACT FROM AUTHOR]

• An innovative GNN-based method for the shear wall layout design that seamlessly integrates design conditions. • The proposed method outperforms others that overlook the importance of embedding design conditions. • The proposed method is versatile and adaptive to arbitrary combinations of design conditions. The construction industry, traditionally labor-intensive, has now been evolving towards automation and the incorporation of intelligence. Notably, the shear wall layout has been a critical component in structural construction, where neural networks have promoted the emergence of sophisticated design methods. These methods integrate graph neural networks (GNNs), successfully mitigating the computational resource demands and challenges in capturing the topological features, which are the impediments in pixel image-based methods. However, the existing GNN-based methods marginally accommodate structural design conditions and underperform in fulfilling practical engineering requirements. Specifically, these methods overlook influential factors such as the peak ground acceleration (PGA) of the design basis earthquake (DBE), characteristic ground period, and building height, all of which are crucial to the shear wall layout design. To address this research gap, this study proposes an innovative GNN-based design method that duly incorporates design conditions—including the PGA of the DBE, characteristic ground period, and building height—and rigorously evaluates its advantages over previous approaches. The findings confirm the efficiency and reliability of the proposed design-condition-informed method and highlight its capability to accurately correlate shear wall layouts with design conditions. [ABSTRACT FROM AUTHOR]