Browsing by Author "Aiordăchioaie, Dorel"

The paper presents a general approach for change detection in vibration signals with application in machine health monitoring using advanced signal processing techniques. These techniques mainly include change detection (CD) algorithms, independent component analysis, time-frequency analysis, energy concentration evaluation, as auxiliary tools for feature extraction and analysis of vibration signals. They operate in a reduced dimension space, highlight and enhance the changes in signals. Finally, the paper presents a case study in health status monitoring of an industrial pump.

The chapter promotes methods of feature design, under the hypothesis that the TF images are easy to compute, i.e. a high capacity of processing data is available. The next section considers the basic aspects of signal processing with data transforms, i.e. time-frequency and cosine transforms. From variously available time-frequency transforms (TFT), the Choi-Williams (CWT) transform is used, based on its good mathematical properties and fast algorithms of computation. Cosine Transform (CT) is also used based on his power to concentrate the information in its coefficients. The results of the experiments based on computer simulation using real data records are presented. Finally, is presented a synthesis of the main results in a qualitative way and some future steps to consider.