Research On Aero-engine Bearing Fault Diagnosis Algorithm And Fault Diagnosis Software Design

As a rotating part of aero-engine,rolling bearing plays a vital role in the smooth operation of aero-engine.Due to the long-term high-speed rotation of the aero engine,as well as high temperature and high load,the rolling bearing is one of the parts with the highest frequency of failures in the aero engine.Therefore,testing and predictive maintenance of aero-engine bearings can effectively reduce the accident rate and avoid property losses.Existing bearing fault diagnosis methods are only applicable to specific speed and load,and cannot effectively diagnose aviation bearings with a wide speed range(1000rpm-14000rpm)and higher loads.The thesis conducts research on the above issues,and the main results are as follows.1.The thesis takes the main shaft bearing of aero engine as the research object,builds an aero-engine main shaft bearing fault simulation experiment platform and bearing vibration signal acquisition system,and collects the aero-engine bearing under multiple loads in the range of 1000rpm-14000 rpm.Four kinds of typical fault data and normal data,and on this basis,extract the spectral features to make a neural network training data set.The paper constructs a neural network for single-channel data and multi-channel data fusion input,and uses the data set to train the two network models.The two network models have the fault recognition accuracy rate in the entire speed range of the aero-engine bearing.Up to more than 98%.The results show that the proposed method has good prospects for practical application.2.The existing fault diagnosis systems for rolling bearings are implemented based on traditional signal analysis methods,and the fault diagnosis system is incapable of bearing fault diagnosis under arbitrary speed and alternating load conditions.In response to the above problems,the paper designs an aero-engine bearing fault diagnosis system based on the Qt platform.The system successfully deployed a single-channel neural network model and a multi-channel data fusion neural network model,and then realized the intelligent diagnosis of bearing faults.In addition to providing a neural network-based bearing diagnosis method,the system also realizes functions such as bearing parameter display,bearing vibration signal collection,and bearing vibration signal analysis.When collecting bearing vibration signals,a ring buffer and multi-threaded read and write methods are proposed,which effectively solves the problem of slow software response or even jamming at high sampling rates.The test results of the diagnostic system show that the system can realize the collection and analysis of bearing vibration signals,and can correctly identify bearing faults,which provides a technical guarantee for the predictive maintenance of aviation bearings.