ABSTRACT
Signal processing plays an important role in gas turbine diagnostics. In this chapter, we use a special type of median filter called the center weighted idempotent median (CWIM) filter to process gas turbine health signals for improved visualization and analysis. The filter requires only two forward data points. Thus, the CWIM filter is useful for gas turbines where data are available slowly. The filter is described in this chapter and then demonstrated on signals containing both deterioration and sharp trend shifts. A key advantage of this filter is that it does not need a priori knowledge of the noise characteristics of the signal. The idea of using the CWIM filter for gas turbine diagnostics was proposed by Ganguli [37] and is discussed in this chapter.
