Weak Fault Diagnosis Of A Class Of Rotor Systems And Its Application

Rotor bearing is the core component of most rotating machinery equipment.Poor working environment and high-intensity operation will inevitably lead to mechanical failure and damage to the rotor system.Early fault detection can identify the fault and the location of the fault as soon as possible,and intervene in the fault to reduce the loss.Stochastic resonance refers to a phenomenon that weak signals are amplified by weak signals and noises under the synergistic effect of nonlinear systems.It is also an effective method for weak fault detection in recent decades.This method realizes the detection of weak signals by amplifying weak signals with noise.In this paper,the shortcomings of the existing weak fault detection method based on stochastic resonance are improved,and the detection effect of weak signal is improved.Therefore,the main research contents of this paper are as follows:1.A fault detection method based on stochastic resonance of adaptive underdamped Fitz Hugh-Nagumo system is proposed.Firstly,according to the sensitivity of the FHN potential to the weak signal and the influence of the damping factor on the transition rate,the Langevin equation under the weak signal,noise and underdamped FHN potential is proposed.According to the adiabatic approximation theory,small parameter expansion,bistable theory and two-state model theory,the expression of the output signal-to-noise ratio of the system is obtained.By analyzing the influence of FHN system parameters on the signal-to-noise ratio,it is concluded that adjusting the system parameters can affect the occurrence of stochastic resonance.The QPSO algorithm is used to adaptively obtain the best parameter of the underdamped FHN potential for detecting weak signals,and the system under this parameter is used for fault detection.Through simulation and detection of experimental data,this method can effectively detect weak signals.2.A fault detection method based on adaptive stochastic resonance noise reduction and VMD decomposition is proposed.This method is based on the stochastic resonance noise reduction of the original signal and the VMD method can customize the mode decomposition number K to avoid mode aliasing expansion.Firstly,the collected signal is denoised by stochastic resonance method based on underdamped FHN potential.Secondly,the number of mode decomposition K is determined by observing the central frequency.Finally,the denoised signal is decomposed by VMD with mode number K,and the fault frequency is identified by observing the time-frequency domain diagram of each mode.The feasibility of the method is verified by numerical and experimental simulation,and the method is slightly better than the method of stochastic resonance detection of underdamped Fitz Hugh-Nagumo system.3.A fault diagnosis method based on time-delayed feedback stochastic resonance of FHN system is proposed.Considering that the historical information has a great effect on the detection of faults and the extraction of weak signals in the later stage,the time delay term is first introduced into the system with overdamped FHN potential.The output signal-to-noise ratio of the system is derived by the small hysteresis approximation and the two-state model theory,and the relationship between the system parameters and the signal-to-noise ratio is analyzed.Finally,the system parameters are optimized by the intelligent optimization algorithm to obtain the optimal parameters and perform fault diagnosis.Through numerical simulation and actual fault data diagnosis,it is found that this method can be effectively used for harmonic fault detection.