Abstract

The paper presents a new method for filtering signals with use of the Modified Recursive Discrete Fourier Transform. The basic idea of the method is the application of user defined context to the recursive form of the DFT and filtration of data or signals. The context is considered as a pre-defined pattern in frequency domain, like frequency envelope of phenomena, which is being searched for. The method works as tool with scanning amplitude-time-frequency representation in the context of such pattern. In this paper authors present mathematical formulas, example of processing noisy signal with predefined patterns and idea of an intelligent algorithm scheme for intelligent context filtering of signals and data. As the example authors filter by the chosen context a genuine diagnostic signal from a large power turbine, where context is applied as a flIter for extraction of certain properties of a diagnostic signal, to obtain knowledge about damages or phenomenas caused by machine dynamics. Future research and possible applications in other fields of research are also discussed.