Research On Weak Signal Detection Of Bearing Of The Wind Turbine Generator System Based On Stochastic Resonance

Due to the environment of wind turbines and the volatility of wind speed, the operation and maintenance costs of wind turbines grows exponentially with the fan running time. And Because of that the wind turbines often work with the fault-prone and long downtime. Bearing components are the main part of the wind turbine failure, and the downtime caused by bearing component failure accounted for 20%. Therefore, early and accurate detection of the bearing failure of wind turbine and reasonable arrangements for maintenance plan can reduce the maintenance costs and increase power generation time.Wind turbine operation will produce high noise, so the fault signal will be submerged in strong noise. For the weak signal detection in the environment of wind turbine, this paper adopted the stochastic resonance method. Stochastic resonance is different from the traditional detection methods. When the noise is strong and the signal energy is weak, this method will restrain the noise as well as increase the energy of the weak signal. So this method can achieve the better detection of weak signal. The paper mainly carried on the following several aspects of the work:In this paper, it first introduce the theory of stochastic resonance(SR). And in order to further improve the denoising effect, we improved the traditional first-order bistable system model of the SR, using the second-order bistable system model. Then I study the influence between the signal-to-noise(SNR) of stochastic resonance system and the parameters of stochastic resonance system.Secondly, I study the modes and causes of the rolling bearing failure as well as the vibration mechanism of bearing and characteristics of common fault signal. After that we obrained the signal characteristics table of typical bearing fault. Then I study the actual vibration signal of wind turbine bearing and the transmission chain of wind turbine. Finally, I obtained that the second-order bistable stochastic resonance can effectively detect the early weak fault of wind turbine bearing.Thirdly, through wavelet packet decomposition combined with genetic algorithm and stochastic resonance, the algorithm of adaptive multiscale noise tuning of stochastic resonance is designed, which is more effective for incipient fault. And the weak fault detection system is developed with the Lab VIEW on the basis of the algorithm. Finally, through analyzing the simulation results, the stability of algorithm and detection system has been verified. The experiments of early fault of the bearing outer ring and inner ring have done on the wind turbine data system(WTDS). And the results show that the algorithm designed in this paper can better extract the feature of early weak fault signal of wind turbine bearing. So this algorithm can accurately detect early weak fault of the wind turbine bearing.