Abstract: Aiming at complex features in the running process of crack rotor system such as non-linear, un-steady abrupt changes, the information entropy feature extracting and fusion analysis method are used to make the dynamic analysis and parameters recognition method combined together in the crack rotor system. The information entropy feature and transformation rule of crack rotor response parameters in various rotating speeds and different states are analyzed to realize the quantitative description of the response feature of crack rotor; Based on this, the proposed new method is used to analyze the testing signals of crack rotor system, and a wavelet neural network model is established to recognize the different information entropy feature curves of crack rotor and realize the fault diagnosis of rotor system. The validity of the proposed method is shown by experimental result.

Key words: Crack rotor, Fault diagnosis, Information entropy, Quantitative description, State feature, Gurney flaps, H-Darrieus wind turbine, Reynolds number, Self-starting, Static torque, Wind energy , Wind tunnel test