Research On Application Of Weak Feature Extraction Based On The Compound And Coupled Multi-stable Stochastic Resonance

With the continuous development of modern production technology,it is of great significance to improve the real-time monitoring and fault diagnosis technology of mechanical equipment.In the early fault diagnosis of mechanical equipment,affected by the on-site environment,the presence of a large amount of noise makes the useful signal characteristic information weak and signal-to-noise ratio(SNR)is low.It is difficult to obtain accurate equipment status information.Therefore,how to detect the weak fault characteristic signal from the noisy mixed signal and improve the SNR is a hot issue in current research.Stochastic resonance(SR)is a weak signal detection technology based on noise enhancement,it is different from traditional weak signal feature extraction methods,SR can use noise to enhance weak signal energy,which has important theoretical significance and application value for realizing early weak fault diagnosis.First,according to the requirements of SR adiabatic approximation theory,this paper proposes a secondary sampling technique,and it is verified by simulation analysis of small and large frequency signals.Then,since the current bistable,multi-stable and monostable models are the potential function models of most SR.In order to further improve the weak signal processing ability of SR systems,according to the application of SR method to the analysis of weak signal extraction techniques and algorithms,this paper improves the SR model by combining the multi-stable model with the Woods-Saxon model,and proposes a composite multi-stable potential well model.This method uses the output SNR as the fitness function of the particle swarm optimization(PSO).Through signal simulation and experimental verification,compared with the system output of traditional multi-stable and bistable models,this method has a larger SNR and more obvious signal period,which proves that the proposed model has better weak signal detection performance.At last,this paper studies the SR enhancement method and establishes a coupled system composed of two multi-stable models.This method has fewer optimization parameters and higher efficiency than the composite multi-stable SR system.The simulation signal analysis and experimental verification results show that the SR system can obtain better detection results after coupling.In summary,this paper studies the noise-enhanced weak signal detection technology based on SR,and applies the researched SR weak signal detection method to fault detection.This method realizes the detection of the simulation fault signal and the bearing inner ring,outer ring and roller faults,which verifies the reliability of the proposed method.