The Research And Realization On Auto Identification For Shaft Orbits Of Hydro-power Sets

In conditional maintenance of hydropower generator sets, the shape of shaft orbits is an important two dimensions graphic symptom. It can diagnose the state of the sets through identifying the shaft orbits. The paper discusses the dynamic principle of the shaft orbits and method of auto identifying.To identify the shape of the shaft orbits ,it must filter it firstly. Harmonic wavelet that has specific function form is an effective tool to analysis signal of vibration. The paper tries to use harmonic wavelet to filter the shaft orbits, which has confirmed the feasibility and advantage in use harmonic wavelet to filter instability signal. In fact recognition of shaft orbits is equivalents to discrimination planar graphs. The paper uses SVD to set up lower dimension vector space. The method based on SVD can be used to identify the shaft orbits of generator sets because it has invariability. Simulate experiments has proved that the method is effective, briefness and feasible. The paper has stressed the method to identify shaft orbits by moment invariant. From 1962 when the method had been present by M.K.Hu, the moment invariant has been used to discriminate graph and image widely. And it has been one of the important tools in pattern recognition. In paper the invariability of moments invariant on discrete condition is discussed from experiments and theories. It has found that moment would be variable in the rotation and scale invariants condition. Improved method has been present and tested through experiments. It has proved that the improved moments have invariability in rotation, translation and scale invariants condition, and it makes characteristic space vials to be classified. The paper describe that noises have effect on invariability of moments and how to chose moments. The method to set up moment invariant when the shaft orbits has been slanting transformed is also present. Finally, the paper presents the hardware framework on monitoring the vibration of hydropower generator sets. And the structure of communication in the system is described.