Real Time Digital Twins for Structural Health Diagnostics

afc90487-1edc-4fec-9e06-644b2c83a191

Closed

EPSRC
EPSRC
EPSRC
EPSRC
EPSRC

No funds listed.

Sept. 30, 2020

Sept. 30, 2023

Detecting damage in engineering structures at the earliest possible time is crucial for almost all private and government industries. In this project computer simulation and testing will be utilised to determine the location of damage in different structures using the Lamb waves. The inclusion of multiple damages in structures and case studies where damages are located in assembled structures (with joints) enables the exploration of various wave propagation techniques and identifying their limitations. One issue with the current approaches is the requirement of large volumes of test data, which are expensive and unlikely to be available. This PhD study considers the use of a validated physics-based model in conjunction with experimental data for the generation of big data sets that will be utilised to develop data-driven damage detection techniques.
The key aims of this project are:

1-To investigate the limits of existing damage detection techniques using wave propagation.
2-To develop digital-twin models capable of generating big data sets for different damage scenarios in structures with multiple damages and joints.
3-To develop data-driven techniques for damage detection using the limited measured data and comparison the performance of these data-driven techniques with existing methods.
The approach to be followed in meeting the above aims includes:

1-Numerical and experimental simulation of wave propagation in aluminium plates with single/multiple cracks and bolted joints. MATLAB/ANSYS will be used for the numerical simulation and standard piezoelectric transducers with high frequency data acquisition will be used for the experiments.
2-MATLAB/Python will be used to generate required data for training the machine learning algorithms for the data-driven techniques.
3-MATLAB/Python will be used to develop data-driven techniques for damage detection.

The novelty of the proposed project lies in using big data generated by digital twin simulations and the development of data-driven techniques for damage detection of structures. The outcome of this project will be the development of data-driven techniques capable of extending the range of applicability of existing methods in the presence of multiple damages and joints in structures. In particular, digital twins will be used to extend the range of available data for structures and damage cases where no data from a physical structure exists. Development of such techniques could potentially save considerable maintenance cost in different engineering structures such as those used in UK nuclear power plants.