The Research On Fault Diagnosis Technology Of Micro-Gas Turbine Rotor

The development of human society depends on electricity,and the development of electricity depends on energy.Nowadays is the period of energy revolution and innovation.In times of energy innovation,distributed energy systems have made outstanding contributions in alleviating the persistent tightness of energy resources and increasing energy demand year by year.Micro-gas turbine is the main core equipment of distributed energy system.Its future research direction is high-power,high speed and high integration.Micro-gas turbine is a branch of the rotating machinery.So there are common failure types of rotating machinery.By consulting the relevant literature,the main fault of rotating machinery is the rotor.But the accuracy for the methods of studying the fault of rotor was not high in the past.And the diagnostic system was occurred some misdiagnosis.Therefore,the research is needed.The following is the content of main research:(1)First,the generating mechanism and fault manifestation of the common fault type is stated for micro-gas turbine rotor.And the main analytical methods are also introduced for relevant faults.(2)Secondly,the main equipment and instruments of the fault data system is introduced.Besides,the wavelet packet decomposition is used to extract and reconstruct the vibration data.(3)Finally,the vibration data is analyzed in the thesis to select the main parameters of BP network for building the BP network fault diagnosis model.Through training and testing,determine the appropriate network model.The innovation of the thesis lies in the fault feature vector,which is determined by the extraction and reconstruction of wavelet packet decomposition,is much easier for BP network input layer to identify.Secondly,after determining the optimal range of the hidden layer nodes in BP network,the optimal value is obtained by training all the hidden layer nodes in the given range,which improves the final BP network performance.Besides,the various improved BP network models can increase the number and type of networks and improve the contrast.Through training and testing,the network model which is determined by the following two indicators,the number of iterations and the level of accuracy,have the high fault diagnosis ability.Through training and testing,the network model which is determined by the following two indicators,the number of iterations and the level of accuracy,have the high fault diagnosis ability on the micro-gas turbine rotor.After verification,by comparing with the standard BP network model,the improved BP network model has more superior performance in terms of running speed and fault diagnosis accuracy.