Deep Learning-based delamination assessment of complex composite structures from UGW responses under varying environment

The present research proposal aims towards developing a Deep Learning (DL)-based inverse delamination damage assessmentapproach in complex industrial composite structures from Ultrasonic Guided Wave (UGW) responses under extreme and varyingoperating and environmental conditions (temperature, humidity, pressure).

The proposal consists of a number of importantinnovative components, such as a) Developing an efficient model for easy incorporation of single and multiple interface delaminationb) Utilizing a mesh-free method to overcome the drawbacks of finite element method c) Modelling accurate wave-damageinteraction under extreme and varying environments d) Constructing a DL-based robust inverse approach to perform effectivelyunder varying structural complexity and operating environments e) Validating the approach for real-life/ laboratory samples.

Meshfree models will provide sufficient flexibility to model geometric complexity and damages besides significant reduction incomputational cost.

DL's capability of handling large data sets and predicting optimum output from raw response will provide asuperior approach to predict damages from raw UGW responses.

Therefore, this proposal will pave pathways to develop the nextgeneration of ‘online’, fast and robust delamination assessment tools for industrial complex composite structures under varyingoperating environments.

This will further enhance European industrial competitiveness and leadership through reducing theinspection cost by assessing the structural integrity of a complex structure without stopping its normal operations.

The Fellow'sexpertise in delamination modelling and assessment and the Supervisor's expertise in modelling UGW propagation in complexstructures will create two-way knowledge transfer between them, which will create major scientific, social and economicadvancement in European aviation, energy and civil industries by providing online and accurate diagnostic and prognostictechnologies.