Research On Condition Monitoring And Fault Diagnosis System Of Bearing In Wind Turbine

Because wind energy is an inexhaustible and less polluted clean energy, it gets rapid development from all countries in the world. With its development, the maintenance work of wind turbine is very important. Once wind turbine has failure or accident, it will lead to great economic loss. In order to ensure that wind turbine operates safely and stably, and reduce the incidence of failure or accident. The condition monitoring and fault diagnosis of wind turbine is needed to be studied.The bearing plays a very important role in the drive system of wind turbine and its work environment is very poor. So the failure occurrence rate of the bearing is higher. According to it, this thesis studies condition monitoring and fault diagnosis of the bearing. Firstly, this thesis introduces the mechanical structure principle of the wind turbine. Secondly, it expounds several typical failure of the bearing and vibration frequency when the bearing goes wrong. Thirdly, it introduces the system design about condition monitoring and fault diagnosis of the bearing. The system includes hardware part and software part. The hardware part includes the construction of the experiment platform, choice of sensor and choice of data acquisition card. The software part makes use of virtual instrument technology and Lab VIEW to complete network communication, feature extraction of signal, fault diagnosis, database management and other functions. Fourthly, this thesis studies the method of wavelet packet de-noising and decomposition, and makes use of time domain analysis, frequency domain analysis and wavelet packet analysis to complete feature extraction of the bearing in the condition of normal, inner race fault, outer race fault and cage fault. Finally, according to the shortcomings of genetic algorithm, this thesis puts forward an improved genetic algorithm. It uses the improved algorithm to optimize the weights of the input layer and the thresholds of the hidden layer of extreme learning machine, and sets up a model of IGA-ELM. Then it applies the model to fault diagnosis of the bearing. The result of the experiment shows that the model gets a good result.