Modified hilbert huang transform for data analysis and its application to vibration signals

Abstract

The Hilbert Huang Transformation, which is an empirical method currently and known as an efficient method in analysis of non-stationary and non-linear signals due to its data-driven nature, constitutes the basis for this thesis. Within the thesis:?Attributed fundamental concepts of data analysis are briefly mentioned,?The original Hilbert Huang Transformation is presented with attribution to various studies,?A new algorithm, which concerns modifications in the formulation of the local mean approximation and adjustment of boundary values, is proposed,?Frequency resolution capabilities of the original and modified transformations are presented comparatively over analysis of a deterministic signal,?Vibration signals acquired from an electric motor, which is subject to an experimental accelerated aging process, are analyzed with:oThe Fourier Transform,oThe Continuous Wavelet Transform,oThe Hilbert Huang Transform, andoThe Modified Hilbert Huang Transform.The first and general justification to include the Continuous Wavelet Transformation in this study is, its theoretical completeness. The second and particular one is, the original and published results obtained in analyses of the mentioned vibration signals with the Continuous Wavelet Transform.The major impacts of the proposed modifications in the algorithm are increased resolution in identification of intrinsic frequencies and reduction in number of iterations to identify them, respectively.Planned future work is stated with justifications in the conclusion part of the thesis.