|
Title
|
|
|
|
Parameter optimization for dynamic line scan thermography
| |
|
Author
|
|
|
|
| |
|
Abstract
|
|
|
|
| Performing non-destructive inspections is widely used to evaluate objects for structural integrity. Infrared (IR) thermography is a common technique in the field of non-destructive testing. Both active and passive thermography are used to inspect objects depending on the type of defect to be inspected. Passive thermography is often used to determine the health of objects that generate heat during operation (e.g. a transformer, a motor, circuit boards, ... ). Deviant heating patterns may indicate anomalies in the samples to be inspected. Active thermography on the other hand is based upon upsetting the equilibrium of the object by means of inducing a thermal wave. Active thermography is often used as non-destructive technique because of its advantages in comparison to others such as ultrasound testing, Xray radiography, visual inspections etc. IR thermography is a fast, safe, contactless and full-field measurement technique to perform inspections. It has shown to be reliable in many applications and therefore more and more industrial companies and researchers show interest in exploring the field of thermographic inspections. Some disadvantages of thermography make it hard to implement the technique as a plug and play technique. The main disadvantage is the fact that an expert skilled in the art is needed to optimise the measurement setup and prescribe the needed parameters in order to perform accurate and efficient measurements. Another important drawback of the thermal inspections performed nowadays is the size limit of objects to be inspected. IR thermography is often performed keeping the sample stationary in the field of view (FOV) of the used camera. Research has been performed in order to inspect larger samples using dynamic line scan thermography (DLST), whereby a heat source and camera tandem moves relative to the specimen to be inspected. Performing measurements using DLST not only requires an expert skilled in thermography, but even one who has expertise in dynamic inspections. Since more parameters have an influence on the result, it becomes much harder to predict the optimal setup parameters. This thesis focuses on providing an insight in the influence of the different parameters used in dynamic line scan measurements and using the acquired knowledge to predict an optimal parameter set. First, a concise explanation of thermography will be given, explaining the working principle of the technology and describing different techniques that can be used in order to perform IR inspections. Second, the focus lies on the measurement setup itself. In order to predict the most suitable parameter set, it is necessary to ensure the optimal working of all individual components. Afterwards, different techniques are investigated to perform parameter predictions and finally a novel method is presented to accurately align the images of the translating object. The proposed techniques are validated using industrial applications. In order to enable the optimisation of the parameter set used for dynamic line scan thermography, an optimal measurement setup had to be defined. There exist line heating sources specifically for DLST measurements, but there were still some improvements possible. Therefore a new line heating source was developed using ray-tracing software. The second goal of this thesis was to generate an insight in the correlation between different parameters of a dynamic inspection. This knowledge enables inspectors to predict possible parameters to perform dynamic measurements without the need of trial and error testing for longer periods of time. Getting insight in the working principle requires a lot of data and would therefore require en enormous amount of measurements. This procedure would be too time-consuming and expensive, resulting in seeking refuge in a finite element (FEM) simulation. The simulation resembles the DLST working principle making it possible to generate data relatively fast. Starting from this data, several methods are investigated to predict the optimal parameter set for dynamic line scan thermography measurements. The used techniques vary from established techniques such as response surface methodology to newer techniques such as artificial intelligence and Gaussian processes. The outcome of the different optimisation techniques is validated using experimental measurements. At first the focus was specifically on predicting the optimal parameter set for DLST based for finding one defect with specific properties. However, techniques investigated in later stages of this thesis, also prove interesting to offer an answer to questions where multiple solutions can be found and focus more on supporting the inspector in selecting a suitable parameter set. Finally the captured images have to be aligned in order to be able to perform postprocessing. Existing methods require a synchronisation between the movement of the sample and the frame rate of the thermal camera. The technique in this thesis enables sub pixel shifting without the need of synchronisation. The knowledge obtained in this thesis can offer an insight in the working principle of dynamic line scan thermography and enables the user to define an optimal parameter set for thermal inspections of samples that do not fit the field of view of the camera. |
| |
|
Language
|
|
|
|
English
| |
|
Publication
|
|
|
|
Antwerpen
:
University of Antwerp, Faculty of Applied Engineering
,
2023
| |
|
Volume/pages
|
|
|
|
xii, 136 p.
| |
|
Note
|
|
|
|
:
Steenackers, G. [Supervisor]
:
Maldague, X. [Supervisor]
| |
|
Full text (open access)
|
|
|
|
| |
|