Building asset management is a complex endeavor that involves development, operation, maintenance and disposal of large-scale costly assets that serve one or more significant functions. Recent and continuing developments in Geographical Information Systems (GIS) offer solutions to the significant challenges of integrating and visualizing asset management data, choosing development proposals, cost assessment, risk assessment and maintenance strategies. Furthermore, GIS data is a common element among many types of projects, buildings and infrastructure assets. GIS technologies can therefore have a significant and broad impact. Navigating GIS developments can be difficult and unclear. To this end, this study performs a literature review on state-of-the-art and emerging GIS technologies as they apply to public asset management. The aim is to provide public authorities with a means to understand the potential and the challenges of these GIS technologies in order to support more informed decision making. The main opportunities that these technologies provide to AM are examined. These include data integration, optimization of resource use, risk assessment and improved decision making from reactive to proactive. In addition, a new Word2Vec K-means based keyword gap analysis tool is proposed to aid in the visualization of keywords in the literature corpus by sorting the keywords into meaningful subject focused categories. This study will help make adoption choices of GIS technologies more informed and coherent, which will allow the reduction benefits in costs, energy and environmental impacts to be more easily leveraged.

Extensive segment floating will result in segment dislocation, crack, and leakage, posing significant risks of engineering accidents. It is important to control the segment floating based on adjusting shield operational parameters finely. A knowledge-based intelligence method designed for controlling segment floating is proposed in this study. Leveraging prior knowledge in segment floating, the framework of the intelligence method is constructed. This framework consists of a segment floating prediction model along with two auxiliary models. The segment floating prediction model considers the spatial and temporal characteristics of the shield operational parameters, including the early activation of the shield excavation parameters and the hysteretic nature of tail grouting parameters. The segment floating prediction model is the basis of the knowledge-based intelligence method. A multi-ring optimization strategy is designed to solve the conflict between the optimization results of adjacent rings. The case study shows that the segment floating prediction model has high prediction accuracy due to consideration of the spatial and temporal characteristics of the shield operational parameters. Considering the performance and computation cost, the optimal parameter configuration is figured out.

Addressing conventional reinforced concrete (RC) shear walls’ susceptibility to brittle failure and residual deformation during earthquakes; this study investigates carbon fiber reinforced polymer (CFRP)-RC composites for enhanced seismic resilience. CFRP’s superior strength-to-weight ratio; corrosion resistance; and self-centering potential address post-earthquake reparability challenges. Current knowledge gaps persist in size-effect mechanisms under combined geometric and reinforcement parameters (shear span ratio; horizontal reinforcement ratio; height-to-thickness ratio). Numerical analysis of 28 models evaluates hysteretic behavior; strength degradation patterns; ductility coefficients; and residual deformation characteristics. A refined size-effect model incorporating CFRP’s strain distribution overcomes existing predictive limitations; advancing performance-based design of damage-tolerant structures.

Grouted Bellows Connect Rebar (GBR) technology is critical for ensuring reliable connections in precast concrete components. The bond-slip behaviour, a core metric for assessing GBR connection performance, presents significant complexity, and existing empirical models often fall short in prediction accuracy to meet engineering demands. Addressing this challenge, this study introduces an innovative hybrid model (CNN-LSTM) that integrates convolutional neural networks with long short-term memory networks. Utilizing eight critical parameters, such as grouting strength, reinforcement ultimate strength, and the anchorage length-to-diameter ratio of the reinforcement, the model achieves precise predictions of GBR bond stress. This study systematically collected data from 114 sets of GBR pull-out tests, constructing a dataset comprising 2,272 bond-slip samples for model training and validation. Additionally, 15 GBR independent samples were independently fabricated and multiple samples were extracted to assess the model generalization capability. Experimental results demonstrate that the CNN-LSTM model significantly outperforms traditional empirical models in predicting bond stress and exhibits superior generalization across key metrics, including total energy consumption, maximum bond stress, failure modulus, and residual energy. Parameter importance analysis reveals that grouting strength, reinforcement ultimate strength, and the anchorage length-to-diameter ratio are the most influential factors in bond stress prediction. Building on the CNN-LSTM model predictions, this study establishes an improved empirical model with clear physical significance, offering a reliable computational foundation for engineering applications.

It is witnessed that building information modeling (BIM) technology has shown its capabilities in data integration in the construction industry. Incorporating innovative geotechnical theories into BIM helps to further develop its application potential. In the practice of deep excavation engineering, obtaining accurate soil parameters is the key to preventing deep excavation accidents and reducing construction costs. Aiming at the complexity of soil properties in soft soil deep excavations, an intelligent inversion framework for soil parameters in deep excavations is established by using BIM technology, finite difference method (FDM), and nondominated sorting genetic algorithm II (NSGA-II). Firstly, a building information modeling-numerical simulation (BIM-NS) integrated component is implemented based on a transformation interface, including geometric meshing processing and controlled script automated execution. Then, a back-analysis component based on NSGA-II optimization is attached to the BIM-NS processing to improve the accuracy of soil parameters. Subsequently, a framework of the building information modeling- numerical simulation-machine learning (BIM-NS-ML) integrated technology is established, enabling the usage of optimal soil parameters for automatic deep excavation simulation. Finally, the integrated framework is applied to a subway deep excavation project, which verifies that the proposed intelligent integration framework can accurately identify soil parameters in an efficient manner. The BIM-NS-ML integrated technology significantly improves the efficiency of modeling and calculating. The multi-objective optimization algorithm effectively addresses the problem of parameter complexity in soft soil. In addition, the intuitiveness of parameter inversion results is further enhanced to provide support for construction management and decision-making.

Smart buildings represent a significant trend in the future of the construction industry. The performance of human-computer interaction plays a vital role in achieving this from a human perspective. However, existing human-computer interaction algorithms are often limited to simple commands and fail to meet the complex and diverse needs of users. To address this issue, this paper introduces large language models (LLMs) and AI agents into smart buildings, proposing a general AI agent framework based on the ReAct strategy. The LLM serves as the system’s brain, responsible for reasoning and action planning, while tool calling mechanism puts the LLM’s plans into practice. Through this framework, developers can rely on prompt engineering alone to enable the LLM to interpret user intent accurately, perform appropriate actions, and manage conversation history effectively, without any pre-training or fine-tuning. To examine this framework, an experiment was conducted in a virtual building, which showed that the proposed agent successfully completed 91% of simulated tasks. Additionally, the agent was deployed on a single-board computer to control devices in a model building, demonstrating its effectiveness in the real world. The successful operation of the agent in this environment highlighted the potential applications of the proposed framework using existing IoT systems, providing a new perspective for the upgrading of human-computer interaction systems in smart buildings in the near future.