MATLAB Scripts for Numerical Model Updating
Description
This dataset contains two MATLAB scripts designed for model updating. A functional overview of each script is provided below: input.mat: Serves as the input layer for training the surrogate model. The input parameters encompass simulation results, experimental results, and update parameter values from both the current stage and the previous stage. output.mat: Serves as the output layer for training the surrogate model. The output parameters consist of the updated simulation results and the effectiveness of the update. NeuralNetwork.m: This function represents the trained surrogate model. Its input and output parameters correspond to the 'input.mat' and 'output.mat' files, respectively. When using this script, a main program can be pre-written to obtain part of the input parameters, and then combined with optimization algorithms to iteratively solve for the update parameters. The update parameters are selected based on the effectiveness of the output parameters. Optimization.m: This script transforms the model updating problem into a multi-objective parameter optimization problem using a genetic algorithm. It enables operations such as modifying, calibrating, running, and extracting results from the Finite Element (FE) model. It can be used to assess the twin fidelity between the real-time digital model and the physical model, thereby completing the model update.
Files
Steps to reproduce
How to Use: 1. Optimization.m can be called within the main program to perform real-time updates of the digital model. 2. NeuralNetwork.m establishes the mapping relationship between update parameters and update effectiveness, and can be used in conjunction with optimization algorithms to achieve model updating.
Institutions
- Southeast UniversityJiangsu, Nanjing
- Aston UniversityBirmingham, Birmingham