Research On Cracked Rotor Fault Characteristics And Identification Of Crack Location And Depth

Rotating machinery has an irreplaceable position in modern industrial production.With the increasing demand for production,high power density and high reliability have become the goals pursued by the development of rotating machinery.Rotors working under severe working conditions such as high temperature,high pressure,and high speed are prone to cracks.Once a crack failure occurs,it may cause a very serious production accident.Therefore,it is of g reat significance to study crack rotor faults and develop crack parameter identification methods.In this paper,the crack element is modeled based on the stress intensity factor method.The strain energy release rate is used to connect the stress and vibr ation of the cracked rotor to the stress location of the crack tip to construct a breathin g crack model;the rotor shaft segment unit is established using the Timoshenko beam theory,and the dynamics of the entire cracked rotor is obtained by assembling th e matrix method equation.Taking the gas turbine rotor as the research object,the vibrati on characteristics of the cracked rotor,the crack depth,the crack position,the size of the unbalance and the phase angle on the subcritical resonance of the crack were studied.Using the time-domain waveform,spectrum characteristics and axis trajectory,the influence of crack depth and position on the vibration characteristics of cracked rotor is summarized.The study found that the 2X component at subcritical speed can be used as the vibration characteristic of cracked rotor.Based on the Kriging surrogate model,a Kriging surrogate model for a cracked rotor is established.The influence of different signal components on the establishment of surrogate model was evaluated.Optimize the crack parameter identification method to reduce the number of measurement points t hat the algorithm depends on the basis of improved accuracy.At the same time,the BP neural network is used to establish the crack rotor neural network model to identify the crack parameters.The advantages and disadvantages of the Kriging surrogate model and the neural network in this thesis are compared in terms of the accuracy of crack parameter identification and the number of samples that the algorithm depends on.