Rolling Bearing Fault Diagnosis Technology Based On Improved Stochastic Resonance Method

Rotating machinery is widely used in modern mechanical equipment,such as steam turbines,compressors,wind turbines,motors,hydraulic turbines,aircraft engine,and other equipment are mainly composed of rotating components.As a supporting part of rotating machinery,rolling bearing almost covers all industrial fields.High speed,variable load,and poor sealing often induce different bearing faults.These local faults will lead to equipment precision reduction,production quality degradation,production interruption,and even system paralysis.Therefore,it is meaningful to realize the fault diagnosis for rolling bearings.The vibration signal characterizing the running state of the rolling bearing often interferes with noise.The strong noise background interferes with the signal analysis and seriously affects the diagnostic efficacy.Therefore,how to efficiently extract the weak fault signal features and improve the signal-to-noise ratio is of great significance.To improve the signal-to-noise ratio,some traditional methods usually adopt the noise suppression that may attenuate a useful signal,which results in a reduced output signal-to-noise ratio and even an anamorphic waveform.This work uses the nonlinear characteristics of the stochastic resonance system to enhance the weak bearing fault signal and extract fault feature information.For these reasons,this work uses the improved stochastic resonance method to extract the weak fault features and realize fault diagnosis for rolling bearings in the strong noise background.The main contents are as follows:1.Based on the analysis of existing re-scale methods,a method named the general scale transformation stochastic resonance is proposed and analyzed in depth in theory.The method does not need to change the target signal,but only needs to find an appropriate system with larger parameters to match a target signal.For a target signal with any frequency,the proposed method can always match signal amplitude with an optimal potential barrier and achieve optimal output.The experimental results show that the general scale transformation stochastic resonance method can accurately extract the weak signal characteristics and greatly improve the output signal-to-noise ratio under the strong noise background.Compared with the normalized scale transformation stochastic resonance technique,the proposed method has obvious advantages.2.Considering the complexity of engineering background noise,Poisson white noise is used to replace Gaussian white noise used in traditional fault diagnosis researches.We studied the general re-scaled theory of stochastic resonance system excited by Poisson white noise.The influences of Poisson white noise parameters and scale coefficient on the stochastic resonance responses are further analyzed.Vibration signals with different local faults are adapted to verify the correctness of general scale transformation stochastic resonance excited by Poisson white noise,and the weak fault features of the bearing are successfully extracted.3.The failure type and fault frequency are usually unknown for the defective bearing.Based on the general scale transformation stochastic resonance theory,this work studies the unknown fault frequency search and fault identification.First,the effects of classic amplitude domain indices to quantify the stochastic resonance system response are analyzed.Then,the search algorithm for unknown fault frequency is described in detail.Finally,the accuracy and efficiency of the algorithm are verified by fault signals under different conditions.The results show that the proposed search algorithm can accurately extract unknown fault features and realize fault diagnosis,and has a good effect on bearing vibration signals with different fault degrees.4.The collected vibration signals often contain a lot of noise,which seriously interferes with the recognition of weak features.This paper further analyses the recovery of vibration signals of rolling bearings with unknown faults and the extraction of weak feature information.In the numerical simulation of unknown signal recovery,according to the characteristics of stochastic resonance response,the amplitude domain index of the piecewise mean value is constructed,and its performance is analyzed in depth.The parameter estimation strategy for harmonic signal,aperiodic signal,and LFM signal are designed.The method of unknown signal recovery under strong noise background is proposed and verified by harmonic signal,binary signal,and LFM signal.The piecewise mean value index is introduced into bearing weak feature extraction,and the effect of piecewise mean value index is analyzed to evaluate stochastic resonance.The ensemble empirical mode decomposition is used to process bearing vibration signals,and the optimal intrinsic mode function is selected based on rotation frequency cut-off criterion and spectral amplification factor.Based on the piecewise mean value index,the adaptive stochastic resonance successfully reproduces the vibration signal,extracts the unknown fault characteristic frequency,and realizes the fault diagnosis.5.We summarized the work and made a prospect for the related researches.In summary,the improved stochastic resonance method is used to extract weak fault features in a strong noise background.The whole paper can be roughly summarized as follows: from Gaussian white noise background to Poisson white noise background closing to engineering practice;from known fault feature extraction to unknown fault diagnosis;from weak signal enhancement to unknown signal recovery.The method proposed in this paper has the advantages of good effect,high efficiency,high accuracy,and easy realization,and has been verified in the rolling bearing fault diagnosis experiment.