Doctoral Dissertations

Date of Award

12-1997

Degree Type

Dissertation

Degree Name

Doctor of Philosophy

Major

Nuclear Engineering

Major Professor

Belle R. Upadhyaya

Committee Members

J. Wasserman, R. E. Uhrig, L. F. Miller

Abstract

The purpose of this dissertation is to develop a systematic approach for fault monitoring and diagnosis for nuclear power plant systems and rotating machinery using short-time Fourier transform (STFT) and wavelet transform techniques during stationary and transient operating conditions.

This research explored the significant improvements in the signal-to-noise ratio when the original signal was decomposed into different levels using the multiresolution analysis (MRA) technique. The MRA was combined with the standard digital signal processing techniques for diagnostics purposes. A data analysis system, that integrates several MATLAB signal processing tools, was developed and implemented with applications to a commercial pressurized water reactor and a rotating machinery system. A method for establishing optimal wavelet selection, using minimum entropy approach, was developed and applied to actual data.

The problems to be investigated in nuclear power plant systems were concerned with the detection and characterization of transients in the data. The stationary or time- dependent characteristics of fuel channel vibration and the estimation of the frequency characteristics of the transient signals were analyzed using the multiresolution analysis. The reactor data analysis revealed the dominant frequencies triggering the dips and spikes in the transmitter's output signals. The coherence, power spectrum and cross spectrum were estimated for the neutron detector and process signals.

The rotating machinery condition monitoring problem addressed the characterization of the start-up and shut-down signatures of a rotor rig under different operating conditions. The prominent spectral components of the signal were tracked in the time-frequency and time-scale domains during rotor speed variations. The combination of STFT and wavelet transform techniques provided a robust approach for detecting incipient changes in the process signal during steady-state operation. These changes might be indicative of possible anomalies in the plant equipment or in the process sensors.

Optimal wavelet functions were established for the problems of interest which were broadly classified as low frequency (0-20 Hz reactor signal) and medium frequency (20-500 Hz motor rig signals) information ranges. The result of this research was an on- line condition monitoring system that integrates multiresolution analysis (MRA) using wavelet transform and the short-time Fourier transform (STFT) method. The signal analysis methodologies were implemented using the MATLAB software tools.

Download
Files over 3MB may be slow to open. For best results, right-click and select "save as..."

Share

COinS