The correlation analysis between current surface cracks of structures and external loads can provide important insights into determining the structural residual bearing capacity. The classical regression assessment method based on experimental data not only relies on costly structure experiments; it also lacks interpretability. Therefore, a novel load estimation method for RC beams, based on correlation analysis between detected crack images and strain contour plots calculated by FEM, is proposed. The distinct discrepancies between crack images and strain contour figures, coupled with the stochastic nature of actual crack distributions, pose considerable challenges for load estimation tasks. Therefore, a new correlation index model is initially introduced to quantify the correlation between the two types of images in the proposed method. Subsequently, a deep neural network (DNN) is trained as a FEM surrogate model to quickly predict the structural strain response by considering material uncertainties. Ultimately, the range of the optimal load level and its confidence interval are determined via statistical analysis of the load estimations under different random fields. The validation results of RC beams under four-point bending loads show that the proposed algorithm can quickly estimate load levels based on numerical simulation results, and the mean absolute percentage error (MAPE) for load estimation based solely on a single measured structural crack image is 20.68%.

Traditional construction progress management is hindered by reliance on manual monitoring, delayed information feedback, and a lack of proactive correction capabilities. To address these issues, this study proposes a data-driven closed-loop framework for dynamic progress management leveraging Building Information Modeling (BIM) and Digital Twin (DT). The proposed framework is operationalized through three integrated modules: (i) a dynamic perception layer that synchronizes on-site conditions via IoT and digitized construction logs; (ii) a stochastic prediction engine coupling machine learning with Monte Carlo Simulation (MCS) to quantify delay risks; and (iii) an optimization module based on a Constraint Satisfaction Problem (CSP) model for automated strategy generation. The system’s efficacy was preliminarily evaluated through a prototype application on a primary school building project. Findings from this case indicate that the framework enables near real-time synchronization for schedule deviation warnings, effectively compressing the information latency from days to within a single management cycle. Furthermore, within the empirical scope of the case study, the implementation of DT-driven strategies was associated with a 3-day schedule advancement relative to the simulated baseline and a 15% reduction in the resource idle rate compared to the pre-deployment phase. This study provides a potential pathway for enhancing progress control precision in similar construction environments.

Real-time mechanical analysis within digital twin (DT) systems requires high-fidelity models that synchronize with the “as-built” state of physical structures. This paper proposes a technical framework for constructing a “mechanical-core” DT by integrating computer vision (CV) sensing with automated finite element model (FEM) updating. Utilizing the Midas API, we developed a platform that automates data acquisition, modeling, and parameter inversion. A momentum-based optimization algorithm is implemented to invert the instantaneous elastic modulus of bridge segments during cantilever construction. The system was validated through a case study of a continuous box-girder bridge. Quantitative results indicate that the initial theoretical model, based on design-phase assumptions, exhibited a mean relative error of approximately 21.9% in vertical displacement. Following the real-time parameter inversion, this error was significantly reduced to less than 0.2% across all monitored construction stages. The rapid convergence (typically within three iterations) and the substantial increase in predictive accuracy demonstrate that the proposed framework effectively bridges the gap between raw sensing data and structural analysis, providing a reliable basis for proactive engineering decision-making.

The preliminary assessment of structural status in reinforced concrete (RC) using visual indicators like surface cracks serves as the primary step in formulating maintenance and reinforcement strategies. To enhance the efficiency of load identification and damage assessment, this study proposes a novel method for determining external load levels on RC beams using structural surface crack distribution images. First, crack distribution characteristics are extracted using image segmentation techniques. Subsequently, mechanical responses of the beam under different load levels are acquired through the finite element method (FEM). Then, this study develops a novel correlation index model by analyzing the relationships between crack distribution images and strain distribution images from the FEM, enabling accurate identification of the load level that best matches the actual crack distribution. Finally, a preliminary assessment of the damage state is conducted through elastoplastic analysis of the RC beam under the optimal load level. Verification analysis based on multiple experimental beam datasets under different load levels demonstrates that the mean absolute percentage error of the method is 10.98%, and the damage assessment results are in good agreement with the crack distribution images.