Abstract:Boundary conditions are crucial for the construction of physical models of concrete-filled steel tube (CFST) arches, but the current model updating methods seldom consider boundary conditions and usually necessitate repeated modifications and recalculations of finite element models, which are not efficient enough. Moreover, the current physics-informed neural network (PINN) provides a new approach for solving physical equations through physical constraint merging into the neural network, but it is still limited by complex structures, multiple scales, and unknown physical model engineering problems. Therefore, this paper proposed a boundary recognition method for physically-enhanced machine learning. First, the basic variables of the physical equation and the variables to be optimized were taken as inputs to enhance the neural network’s ability to predict the target response. Then, similar to the handling of physical loss constraints in the PINN, the physical loss between the physical model (FE) and the neural network predictions was used to update the parameters. Then, a physical feedback-enhanced neural network (PFENN) meta-model was established, and modified mutation particle swarm optimization (MMPSO) was adopted to optimize the identification of boundary parameters. The identification process involves validation via mutual verification using a finite element model, meta-model, and experimental model. The prediction accuracy is higher than that of classical Kriging, response surface (RS), support vector machine regression (SVR), and neural network (NN) meta-models; additionally, the proposed model has good optimization stability, and reliable optimization results, and its computational efficiency is increased by 4.57 times compared with that of the traditional non-metamodel method. Moreover, the proposed method enhances the accuracy and interpretability of surrogate models by enhancing physical parameters and reducing physical losses, which is suitable for boundary recognition and physical model construction of different structures.

Method for identifying boundary conditions of CFST arches using physically enhanced machine learning

DF-CDM: Conditional Diffusion Model with Data Fusion for Structural Dynamic Response Reconstruction

High-Efficiency Finite Element Model Updating of Bridge Structure Using a Novel Physics-Guided Neural Network

Structural damage identification via physics-guided machine learning: a methodology integrating pattern recognition with finite element model updating

CFRP-RC fracture strength prediction: A novel approach using deep learning

Multiscale Analysis of the Arch Feet Zone for CFST Arch Bridges with Steel Truss and Tie-Beams

Structural mechanism-based intelligent capacity prediction methods for concrete-encased CFST columns

Prediction of ultimate conditions and stress–strain behaviour of steel-confined ultra-high-performance concrete using sequential deep feed-forward neural network modelling strategy

Civil Engineering Structural Damage Identification by Integrating Benchmark Numerical Model and PCNN Network

DNN surrogate model based cable force optimization for cantilever erection construction of large span arch bridge with concrete filled steel tube

Prediction and Global Sensitivity Analysis of Long-Term Deflections in Reinforced Concrete Flexural Structures Using Surrogate Models

Ultimate Conditions Prediction and Stress–Strain Model for FRP-Confined Concrete Using Machine Learning

Hybrid structural analysis integrating physical model and continuous‐time state‐space neural network model

OPERATIONAL MODEL IDENTIFICATION AND CALIBRATION OF CFST ARCH BRIDGE BASED ON AMBIENT EXCITATION

Satin bowerbird optimizer-neural network for approximating the capacity of CFST columns under compression

Intelligent prediction methods for N–M interaction of CFST under eccentric compression

Physics-guided deep neural network for structural damage identification

Physics-informed machine learning model for computational fracture of quasi-brittle materials without labelled data

Surrogate Neural Network Model for Prediction of Load-Bearing Capacity of CFSS Members Considering Loading Eccentricity

Towards a Hybrid Digital Twin: Physics-Informed Neural Networks as Surrogate Model of a Reinforced Concrete Beam

Efficient Model Updating of a Prefabricated Tall Building by a DNN Method