Research On Weak Fault Diagnosis Method For Subway Gear Box Bearings

As one of the most key components of subway vehicles,its operational safety condition plays a vital role in healthy work performance of subway vehicle gearbox.The research and development of the fault diagnosis algorithm for the subway gear box bearing has great theoretical and realistic value.But the gear box bearing has a lot relative components and most of them work in harsh conditions,which makes the fault feature signals for fault diagnosis are easily annihilated in the rest of the useless components or noise.Signals obtained under such conditions become weak fault signals,and cannot be diagnosed by traditional time domain analysis or frequency domain an alysis.Therefore,it is important to study weak fault diagnosis method of the subway gear box bearing to ensure the orderly operation of the subway gearbox.Aiming at the characteristics of strong noise and hidden fault features of vibration signal of subway bearing,author studied the denoising algorithm and proposed an improved fault diagnosis method based on blind source separation and stochastic resonance.Based on the simulation signal generated from dynamic model,the effectiveness for simulation signals of the proposed algorithm has been verified;based on the actual signal obtained from metro gear box fault simulation test bench,the effectiveness for actual signals of the proposed algorithm has been verified.(1)Based on the characteristics of ingredients complexity and strong noise in the vibration signal of subway gear box bearing,this paper studied the traditional TDA and WPA denoising algorism.On this basis,this paper proposed an improved TDA and WPA denoising algorism.And the energy transfer characteristic of the algorithm is verified by the measured vibration signal.(2)Aiming at the characteristics of the difficult vibration source signal separation in single-channel vibration signal in the fault subway gear box bearing.This paper proposed a single-channel bearing weak fault feature extraction algorism based on improved EMD-BSS used for subway gearbox.With EMD algorism,this proposed algorism can enhance the dimension of the single-channel vibration signal.And with the calculation the correlation matrix of the multi-IMF components,the signal can be reconst ructed with intentional choose.Then conducted blind source separation on reconstructed signals based on Cardoso JADE.Based on both simulation signals and actual signals,the effectiveness of the proposed algorithm has been verified.(3)For the weak fault signal is easy to be annihilated by noise,the stochastic resonance method is used to enhance the fault feature energy.After the parameter optimization realized by artificial fish swarm algorithm,the large parameter stochastic resonance system can be transformed into small parameter system which is satisfied with the adiabatic approximation condition.So the weak signal enhancement can be achieved,and the improved EMD-BSS algorithm can be used based on Fast ICA BSS.At last,the Hilbert envelope is used to analyze the signal process results.The validity of theproposed algorithm is verified by the simulation signals and measured signals.In this paper,subway vehicle gear box bearing is the research object,and weak fault diagnosis algorithm for subway vehicle gear box bearing is the study goal.This paper studied the subway bearing vibration mechanism and failure bearing vibration signal simulation.The development of a weak fault signal diagnosis algorithm for subway bearings cannot only detect the generation of weak fault,but also can change the existing maintenance mode which based on operational time,so as to enhance the running safety of subway vehicles while reducing operating costs,then improve the level of operation and maintenance of subway industry and the entire rail transport industry.