Optimization of wavelet bases of EMG signal for application in design and dynamics of machines and mechanisms
Loading...
Date
Authors
Τζήμας, Ιάσων
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
In the following thesis an investigation is carried out towards the direction of obtaining optimal Wavelet bases, in a sparsity sense for signals commonly found in biomedical engineering applications. Most of the inspiration that lead to this work was drawn from studying Ilizarov equipment and the possible scenario of incorporating clever signal processing into properly adjusting the parameters of the device through adequate actuators. Our world is packed with data that are in nature immensely complicated, containing rich, hidden structure. For these reasons the extraction of useful information is oftentimes a lasting and exhausting
process. Mallat aptly describes this, as trying to find a needle in a stack of hay. It is
thus of great usefulness to identify Dictionaries of sets of Vectors that lead to meaningful and intuitive decompositions of the signal. Additionally, digging deep into the structural elements of biomedical data can give intuitive feedback on how mammals use their muscle and bone compositions for ergonomic, efficient locomotive tasks. This information can then be used to push the limits of machine design and automatic control even further. The first section provides an overview of the main key areas of interest with extended reference to their past and present ,as well as their relation to the discussed topic and their state of the
art. Introduction is followed by the theoretical background that was needed to construct the algorithms and the code scripts used in this thesis to derive the results and the answers to the main questions that were posed.
The two approaches for the identification of optimal Wavelet bases are explained and presented both algorithmically and theoretically in section 3. Emphasis is given at the sparsity metric of the Wavelet representations, at the programming resources management, computation time and at cross-validating the results with a wide range of signals obtained from varying subjects. The results of the implementations of the two approaches mentioned above are presented
in section 4. The validity and adequacy of the results is examined and a comparative study is conducted between the identified Waveforms and the preexisting state of the art ones.
Description
Keywords
Biomedical signals, Wavelets