Research And Implementation Of DFIG Rotor And Stator Fault Diagnosis Based On Improved HHT

With the rapid development of wind power globally and the rapidmature of wind power technologies, large and even super-large capacity ofwind turbines are being developed, produced and used increasingly. As thecore part of wind turbine, generator’s safe and reliable operation is directlyrelated to the security and stability of the whole wind turbine, the powerquality and even the staff’s safety. Studies show that the rate of stator androtor faults of double feed induction generator(DFIG) is nearly50%, thusit’s of certain significance to do research about the fault diagnosis of DFIG.This paper’s contents are as follows:Firstly, different stator and rotor fault diagnosis methods and failuremechanisms are detailed. After those researches, Motor Current SignalAnalysis (MCSA) method is chosen to diagnose stator and rotor faults.Secondly, the traditional nonlinear and unsteady signal processingmethods are summarized and analyzed. The advantages and disadvantagesof using those methods are also been given. Then we recommend using anewer method named HHT to process nonlinear and unsteady signals. Thisalgorithm has a good adaptability, which shows the time-frequency relationof a signal clearly. It is useful and important for us to diagnose DFIG statorand rotor faults. But two troublesome problems in application of HHT areend effect problem and false intrinsic mode functions. An improved solutionmethod based on adaptive waveform matching endpoint extension isproposed to solve end effect problem. Meanwhile, an improved energyconservation method is proposed to eliminate false intrinsic mode functions,so we can improve the precision of HHT when using it to diagnose DFIG faults.Thirdly, on the one hand, by using STM32F103MCU and hall currentsensor, we made a data acquisition card. The current datum are directlytransmitted to PC via USB port, and the datum processing are done by a LabVIEW program. The Lab VIEW program calls matlab codes to analyzecurrent datum of DFIG. The IMF result, Hilbert spectrum, Hilbert marginalspectrum and FFT spectrum will be given for engineers. On the other hand,we made a portable DFIG stator and rotor faults diagnose analyzer by usingthe embedded Linux system and QT program on the S3C2440ARMplatform. The current datum are transmitted to the portable analyzer viaUART port. The datum processing are done by improved HHT algorithm,and the Hilbert marginal spectrum and FFT spectrum can be calculated forengineers. Based on the analysis of the fault characteristic frequency inthese spectrums, we can determine the faults happen or not and theirseverity.Finally, by controlling the speed of a DC motor, which is connected to aDFIG, simulate an actual wind turbine operating environment. We havereformed the stator and rotor windings, so we can simulate stator windingsturn-to-turn short faults, stator winding phase-to-phase short faults; threephase unbalanced faults of stator and rotor windings. By using improvedHHT algorithm to analyze the current waveform of those faults under50%load and rated load conditions, we can diagnose the DFIG stator and rotorfaults effectively. For the actual DFIG stator and rotor fault diagnosis, thisresearch results also have a good guiding significance and application value.