Rolling Bearing Weak Fault Detection And Intelligent Condition Assessment Under Complex Operating Conditions

In the operation environment of high temperature and high load,aero-engine rolling bearing failure occurs frequently.Such occurrence may lead to a catastrophic flight accident.Hence,real-time monitoring the aero-engine rolling bearing operating condition as well as trying to detect the bearing initial fault as early as possible is of great significance to flight safety.In this paper,weak fault detection and condition intelligence evaluation methods of the rolling bearing under complex operating conditions are studied.In this study,there are some discussions as follows:(1)The mechanism of vibration generation and fault signal characteristics of rolling bearing are analyzed.According to the characteristics of rolling bearing signal of bearing surface fault,the simulation models of rolling bearing outer ring fault signal,inner ring fault signal and ball fault signal are carried out.(2)Based on the simulation signal,the performances of 3 signal processing methods are analyzed,including the minimum entropy deconvolution,the wavelet decomposition and the autocorrelation analysis.On this basis,a multi-method cooperative diagnosis technique for rolling bearing fault diagnosis is proposed.Meanwhile,proposed cooperative diagnosis technique is popularized to the condition of the speed time-varying by applying the angle domain resampling technique.Plenty of have been rolling bearing fault experiments,in which bearing faults include single point fault of inner raceway,outer raceway and ball surface generated by artificial WEDM,as well as natural stripping fault of inner raceway caused by fatigue test,have been carried out under different operation conditions by applying the rotor tester with a casing while the vibration acceleration signal were measured from the casing during the experiments for method verification.Such experimental results proved that the proposed method can accurately realize the rolling bearing fault diagnosis under the weak fault signal.(3)Based on error analysis,a novel algorithm for feature vector spatial distribution hyper sphere optimization is proposed and a novel data description method called hyper-spherical distance discrimination is proposed.The effectiveness of the proposed method is fully verified by the bearing single point fault experiments.Then,the proposed hyper-spherical distance discrimination method was applied to the rolling bearing condition assessment.The rolling bearing fatigue experiment and aero-engine bearing failure acceleration experiments have been carried out and the experimental results showed that the evaluating indicator obtained by hyper-spherical distance discrimination is more sensitive to rolling bearing initial fault than the root mean square.