Measurement of LSP-induced residual stresses and fatigue life on thin-walled aluminium alloys specimens

This thesis work encloses activities carried out in the Laser Center of the Polytechnic University of Madrid and the laboratories of the University of Bologna in Forlì. This thesis focuses on the superficial mechanical treatment for metallic materials called Laser Shock Peening (LSP). This process is a surface enhancement treatment which induces a significant layer of beneficial compressive residual stresses underneath the surface of metal components in order to improve the detrimental effects of the crack growth behavior rate in it. The innovation aspect of this work is the LSP application to specimens with extremely low thickness. In particular, after a bibliographic study and comparison with the main treatments used for the same purposes, this work analyzes the physics of the operation of a laser, its interaction with the surface of the material and the generation of the surface residual stresses which are fundamentals to obtain the LSP benefits. In particular this thesis work regards the application of this treatment to some Al2024-T351 specimens with low thickness. Among the improvements that can be obtained performing this operation, the most important in the aeronautic field is the fatigue life improvement of the treated components. As demonstrated in this work, a well-done LSP treatment can slow down the progress of the defects in the material that could lead to sudden failure of the structure. A part of this thesis is the simulation of this phenomenon using the program AFGROW, with which have been analyzed different geometric configurations of the treatment, verifying which was better for large panels of typical aeronautical interest. The core of the LSP process are the residual stresses that are induced on the material by the interaction with the laser light, these can be simulated with the finite elements but it is essential to verify and measure them experimentally. In the thesis are introduced the main methods for the detection of those stresses, they can be mechanical or by diffraction. In particular, will be described the principles and the detailed realization method of the Hole Drilling measure and an introduction of the X-ray Diffraction; then will be presented the results I obtained with both techniques. In addition to these two measurement techniques will also be introduced Neutron Diffraction method. The last part refers to the experimental tests of the fatigue life of the specimens, with a detailed description of the apparatus and the procedure used from the initial specimen preparation to the fatigue test with the press. Then the obtained results are exposed and discussed.

Deep learning for computer vision: a comparison between convolutional neural networks and hierarchical temporal memories on object recognition tasks

In recent years, Deep Learning techniques have shown to perform well on a large variety of problems both in Computer Vision and Natural Language Processing, reaching and often surpassing the state of the art on many tasks. The rise of deep learning is also revolutionizing the entire field of Machine Learning and Pattern Recognition pushing forward the concepts of automatic feature extraction and unsupervised learning in general. However, despite the strong success both in science and business, deep learning has its own limitations. It is often questioned if such techniques are only some kind of brute-force statistical approaches and if they can only work in the context of High Performance Computing with tons of data. Another important question is whether they are really biologically inspired, as claimed in certain cases, and if they can scale well in terms of "intelligence". The dissertation is focused on trying to answer these key questions in the context of Computer Vision and, in particular, Object Recognition, a task that has been heavily revolutionized by recent advances in the field. Practically speaking, these answers are based on an exhaustive comparison between two, very different, deep learning techniques on the aforementioned task: Convolutional Neural Network (CNN) and Hierarchical Temporal memory (HTM). They stand for two different approaches and points of view within the big hat of deep learning and are the best choices to understand and point out strengths and weaknesses of each of them. CNN is considered one of the most classic and powerful supervised methods used today in machine learning and pattern recognition, especially in object recognition. CNNs are well received and accepted by the scientific community and are already deployed in large corporation like Google and Facebook for solving face recognition and image auto-tagging problems. HTM, on the other hand, is known as a new emerging paradigm and a new meanly-unsupervised method, that is more biologically inspired. It tries to gain more insights from the computational neuroscience community in order to incorporate concepts like time, context and attention during the learning process which are typical of the human brain. In the end, the thesis is supposed to prove that in certain cases, with a lower quantity of data, HTM can outperform CNN.