Example with non-monotonic reference velocity

The same example is reused to simulate a run up followed by a run down. This implies that the reference velocity is not strictly monotonic during the whole test. The RPM goes from 700 to 7000 RPM and decreases to 3000 RPM.

Figure 22: Non-monotonic RPM

As the RPM is used on the X axis of the analyzed curve, the fact that it is not monotonic will create an overlap of the curve. The following figure shows the analyzed curve plotted with respect to the time (top) and to the RPM (bottom).

Figure 23: Analyzed signal v time and RPM

The spectral map applied to the top curve (signal vs time) shows the order curves going backwards after 60 s. This corresponds to the end of the run up and the start of the run down.

Figure 24: Spectral map scaled with time

The spectral map applied to the bottom curve (signal vs RPM) generates a question asking the user whether the process should continue although some FFT slices may overlap. Clicking on Yes forces the use of RPM as the Y axis. Clicking on No will produce a spectral map similar to the previous figure.

Figure 25: Non-monotonic RPM warning

The following figure shows that the generated spectral map seems to have more slices after 3000 RPM. This is caused by slices corresponding to the end of the run up overlapping slices corresponding to the run down.

Our system example fortunately has a symmetrical behavior. However, real-life systems may not behave the same way. You must avoid the run up overlapping the run down to get clearer results1.

Figure 26: Spectral map scaled with non-monotonic RPM

1

"Numerical Recipes -- The Art of Scientific Computing" W.H. Press, S.A. Teukolsky, W.T. Vetterling, B.P. Flannery, Cambridge Ed.