Research On Noise Reduction And Feature Extraction Of Acoustic Emission Signals In Dry Gas Seal

With the development and progress of science and technology and production automation process,dry gas sealing intelligence is an inevitable trend of its future development.The research on real-time monitoring and fault diagnosis of dry gas seal operation status is of great significance.As a non-destructive testing technique,acoustic emission testing has proven to be very sensitive to the tribological behavior of rotating machinery such as turbines,bearings,and mechanical seals,with the potential to detect early failures in dry gas seals.Therefore,this paper builds a lowspeed and high-torque test bench for dry gas seal,and designs a dry gas seal acoustic emission test system by using high-frequency wide area and strong sensitive acoustic emission instrument.The acoustic emission signals in dry gas seal start-stop,pressureless,low-pressure,high-pressure,low-speed,high-speed and several typical fault operating conditions are monitored.The four aspects of dry gas seal acoustic emission signal acquisition,noise reduction and reconstruction,feature extraction,and operation state analysis and identification are studied.Firstly,aiming at the common problem of signal endogenous coupling and external interference factors in the low-speed stage of dry gas seal,a local characteristic-scale decomposition and wavelet threshold co-noise reduction method are proposed,which effectively strips the useless interference signal from the pure signal.Based on the local feature scale decomposition combined with the wavelet threshold method and the information mapping relationship of acoustic emission signal noise reduction,a mechanism for identifying the noise components of the number of correlations is established,and the pure components and the noise components after noise reduction are reconstructed to achieve signal noise reduction,and a mechanical seal acoustic emission test bench is built to collect the dry gas seal acoustic emission signal in real time for test verification.The experimental results show that the noise reduction effect of LCD-new threshold noise reduction method is better than that of LCD forced noise reduction and wavelet threshold noise reduction,and the signal-tonoise ratio of LCD-new threshold noise reduction is 20% and 23% higher than that of LCD forced noise reduction and wavelet threshold noise reduction,respectively.It is proved that the noise reduction technology based on the sealed acoustic emission signal and wavelet threshold under the local feature scale maintains the usability of the signal,ensures the signal fault characteristics,and lays a theoretical foundation for the whole life cycle management of dry gas seal.Secondly,a mechanism for identifying the operating state of dry gas seal with root mean square as the evaluation index is established to analyze the signal and identify the operating state of the acoustic emission signal under the global operation of dry gas seal.Acoustic emission signal signature recognition is carried out for three different friction states existing in the sealing operation,namely boundary lubrication(BL),hybrid lubrication(ML),and hydrodynamic lubrication(HL)state.A special test bench is used to record the dry gas seal AE signal under different operating conditions of the mechanical seal.The test data is then processed using time-domain,frequencydomain,and time-frequency domain analysis methods.The results show that root mean square(RMS)has good identification performance for different states of sealing operation,and the frequency of acoustic emission signals in the contact and grinding state of the sealing end face is concentrated in high frequency(240k Hz～320k Hz).At last,the fault sources of three typical faults of dry gas seal are clarified,the operation status of dry gas seal fault is set,the AE signal under the single fault of three types of dry gas seal is collected,and the mapping relationship between the fault response and signal characteristics of the three types of dry gas seal is explored.The AE signal operating under three typical faults of dry gas seal was studied,and the time domain analysis and frequency domain analysis methods were used to verify the signal frequency band of the mechanical seal during dry friction,and the relationship between AE signal energy and speed of mechanical seal under dry friction,end face defects and spring failure was explored.It lays a theoretical foundation for the research and development and engineering application of new intelligent fault diagnosis methods for mechanical seals.In this paper,the noise reduction and global operation of the acoustic emission signal of dry gas seals and the extraction of acoustic emission signal characteristics of typical fault conditions are mainly studied,and the research results of this paper have laid a solid theoretical foundation for the future development of dry gas seals in the direction of intelligence,high reliability and long life.