Study On Fault Feature Extraction Method Of Large Steam Turbine

The steam turbine is the key equipment of the thermal power generation. It will lead to enormous losses once the turbine generator set is out of control. The vibration sources of the large steam turbine which have nonlinear, non-stationary characteristics are so complicated.The fault signal of steam turbine is simulated. Some common faults of steam turbine are listed. And the time domain waveform, spectrum and axis orbit are described. This work is the foundation of the turbine fault database.Independent Component Analysis (ICA) which is a new method has been widely used in recent years in many industrial areas to decompose mixed signals into several mutual independent components. The frequency domain ICA based on Fourier transform is adopted to analyze the turbine vibration signals. First the observed signals are carried on Fourier transform. Then the transformed signals are separated by ICA. The independent components are achieved by inverse Fourier transform. The turbine vibration signal is analyzed by the frequency domain ICA. The result shows that frequency domain method of ICA can separate the mixed signal clearly and it is stable in regular condition. The separating results of frequency domain ICA are closer to the actual vibration of steam turbine.The single channel ICA is adopted for the problem which the number of observed signals is less than that of separated independent components. The signal is divided into segments, which are separated by ICA, and the basic functions and sources can be get. Single channel ICA is used to analyze the turbine vibration signal in the text. The basic functions of shaft vibration and bearing vibration are contrasted under two different conditions. It can be seen from the separated results that the basic function of bush vibration signal is different in stable and unstable condition. But the difference of separated result of shaft vibration is not obvious.