Spectral estimator for dynamic-test data from turbine-blade vibrations

This paper presents an analysis of the distribution of occurrence times for pulse trains which can arise in the measurement of turbine-blade vibrations. The problem arises where a proximity probe, mounted in the casing, detects the passing of a rotor-blade tip and generates a pulse. Since the pulse timing depends on the magnitude of the blade vibration an irregular distribution of time intervals arises, and here the properties of the pulse-occurrence distribution are related to the blade-vibration statistics. In particular, a spectral estimator is considered and its errors are evaluated. The analysis provides a solution to a general problem in the estimation of the power-spectral-density function of a Gaussian process from observations of crossings at an arbitrary barrier.