Modeling And Detecting Incipient Fault Of Wind Turbines

In recent years, maintenance costs of wind turbines have greatly increased. The traditional way of fault detection and repair has been difficult to fulfill the requirements. Now it is need more sensitive for incipient fault of the wind turbines,so that people can take measures and arrangements to avoid large losses earlier.In order to detect generator incipient faults, such as stator windings inter-turn and broken rotor bars, have suc h characteristics with small amplitude, non-stationary and susceptible to load variations. Sample entropy algorithm is introduced to perform fault feature extraction from the stator current and electromagnetic torque signals of generators in wind turbines. The proposed method is used to analyze fault signals under different load conditions to realize the quantitative representation of faulty signals.Additionaly,a model for the Doubly Fed Induction Generator with stator winding inter-turn short circuit fault has erected based on mathmatical equations, validity of model was verified through simulation experiments,and using generator control loop signals to detect its fault. The specific research ways are as follows:Firstly, the physical model of Squirrel-cage Induction Generator with stator winding turn faults was analyzed in this paper. A simplified fault model in two-phase stationary coordinate system was built by the coordinate transformation, which launched the state equations of induction motors in fault and write the corresponding program.And in the Matlab/Simulink environment,a model of induction generator has simulated for broke n bars by changing the nonuniform distribution of the rotor winding resistance value.Secondly, Sample entropy algorithm is introduced to perform fault feature extraction from the stator current and electromagnetic torque signals of generators in wind turbines. The proposed method is used to analyze fault signals under different load conditions to realize the quantitative representation of faulty signals. The results demonstrate that sample entropy algorithm is applicable to achieve fault characteristics quantification with data of short length, especially under varying operations and noise environments and it has great potentials in early fault detection and real-time online monitoring.Thirdly, this paper erects the model for the Doubly Fed Induction Generator with stator winding inter-turn short circuit fault. The model consists of the following unit module:wind turbine drive system、stator flux observer、doubly fed induction generator body、rotor-side control system.The MPPT control and the decoupling between active and reactive power can be realized by this model. Different degrees of fault simulation for the double-fed wind power generation system with stator-turn short circuit fault.To extract the stator current m t axis signal for analysis and the results demonstrate that this signal can detect doubly fed induction generator with stator winding inter-turn short circuit fault under different variaton wind speed condtions.