Fault Detection And Diagnosis

Fault Detection and Diagnosis

A use of the principle of amplitude modulation that is particularly important in the practice of electromechanical systems, is in the fault detection and diagnosis of rotating machinery.

In this method, modulation is not deliberately introduced, but rather results from the dynamics of the machine. Flaws and faults in a rotating machine are known to produce periodic forcing signals at frequencies higher than, and typically at an integer multiple of, the rotating speed of the machine.

For example, backlash in a gear pair will generate forces at the tooth-meshing frequency (equal to the product: number of teeth× gear rotating speed).

 Flaws in roller bearings can generate forcing signals at frequencies proportional to the rotating speed times the number of rollers in the bearing race. Similarly, blade passing in turbines and compressors, and eccentricity and unbalance in the rotor can generate forcing components at frequencies that are integer multiples of the rotating speed.

The resulting

system response is clearly an amplitude-modulated signal, where the rotating response of the machine modulates the high frequency forcing response.

This can be confirmed experimentally through Fourier analysis (fast Fourier transform or FFT) of the resulting response signals. For a gearbox, for example, it will be noticed that, instead of getting a spectral peak at the gear tooth-meshing frequency, two side bands are produced around that frequency.

Faults can be detected by monitoring the evolution of these side bands. Furthermore, since side bands are the result of modulation of a specific forcing phenomenon (e.g., gear-tooth meshing, bearing-roller hammer, turbine-blade passing, unbalance, eccentricity, misalignment, etc.), one can trace the source of a particular fault (i.e., diagnose the fault) by studying the Fourier spectrum of the measured response.

Amplitude modulation is an integral part of many types of sensors. In these sensors a high-frequency carrier signal (typically the ac excitation in a primary winding) is modulated by the motion.

Actual motion can be detected by demodulation of the output. Examples of sensors that generate modulated outputs are differential transformers (LVDT, RVDT), magnetic-induction proximity sensors, eddy-current proximity sensors, ac tachometers, and strain-gage devices that use ac bridge circuits.

Signal conditioning and transmission would be facilitated by amplitude modulation in these cases. The signal has to be demodulated at the end, for most practical purposes such as analysis and recording.