Research On Integrated Time-Frequency Analysis Methods Of Wind Turbine Drive System Fault Diagnosis

Wind energy, one of the earliest human use of energy,is rich reserved and renewable clean energy, because it does not generate any harmful gas and waste in the conversion to electrical energy, so it does not contribute to the environmental pollution, for whitch it attracts great attention from the governments all over the world.With the rapid development of wind energy technology and it is becoming more and more perfect, the wind power generator gets more and more high reliability, but during the rapid development of wind power system, wind turbine fault problem has received a great deal of attention at the same time,the wind turbine itself bearing wear, gear, shaft eccentricity, can cause the wind power generator destroyed, reducing the economic benefits, in view of these, wind power generator fault diagnosis has become an important research content in the development of wind power generation.Because of working for long time, the friction during the parts of the instrument component can make them aging or attrited, it will undoubtedly reduce the life of generator, reducing the working efficiency of the generator, most common failure are the w gear and bearing attrition,there are many method to diagnise the fault,such as direct observation method, the vibration and noise testing, nondestructive testing, wear residue detection,machine performance parameter detection method.In order to effectively diagnose the fault, direct observation method to judge the traditional cannot satisfy the development of wind turbines and our requirements,one of the most commonly used is the use of vibration detection method, the vibration signal is a carrier of information equipment, including rich fault feature information, through various signal processing method,we can extract,the characteristics of information hiding in the vibration signals, to realize the diagnosis equipment.Using the wind turbine vibration signal to judge the fault is a reliable method, and time-frequency approach to vibration signal is the most appropriate signal analysis method, the traditional time-frequency analysis methods such as short time fourier transform, Wigner-Ville distribution and so on, but there are some problems during them,such as window effect and the cross term problem,what is worse,they are not adaptive.This paper studies a new adaptive time-frequency analysis method Hilbert-Huang transform(HHT), and applicate it in vibration signal processing of wind power generator.Hilbert_Huang transform by the empirical mode decomposition(EMD) method to decompose the signal into several intrinsic mode(IMF),Each intrinsic mode is a steady signal, it can make Hilbert transform, through the Hilbert spectrum of the signal, we can obtaine the signal characteristics, then combined with the fuzzy neural network system, we can more accurate and conveniently identify the type of fault.However, the traditional Hilbert_Huang transform has ends and termination conditions problems, this paper studies a series methods aimed at resolving the two defects,to forme a kind of improved HHT method,by this improved HHT analysis method to obtain the time-frequency characteristics of vibration signal generator, but the HHT just get the time-frequency characteristics of signals, it doesn’t get the fault diagnosis results, in the present study,Fuzzy neural network has been rapid development, and shows a strong advantage, it is a kind of fuzzy logic reasoning and powerful structural knowledge expressing ability and strong self-learning ability of neural network in the integration of technology,During this paper,we will use the HHT time-frequency characteristics combined with fuzzy neural network to diagnose and identify the type of fault, obtained very good result.