Research On Weak Signal Detection Based On Stochastic Resonance And Empirical Mode Decomposition

Weak signal detection technology utilizes physics and signal processing methods to extract weak quantities from noise,which involves life science,image processing,mechanical fault diagnosis and other application fields.In the process of signal detection,the traditional noise reduction technologies can extract the weak signal characteristics,but these technologies weaken the signal energy.Stochastic resonance(SR)is a special nonlinear detection method,which makes use of the interaction among the signal,system and noise to enhance the weak signal energy.It makes the characteristics of weak signal obvious.Therefore,SR has important research value in weak signal detection field.This paper describes the basic theory of SR in details and introduces its important achievements in the weak signal detection field.What's more,on the basis of the existing weak signal detection methods,this thesis studys different weak signal detection methods of stochastic resonance systems from frequency domain and time domain,respectively.The main works of the thesis are as follows:(1)Weak signal detection of stochastic resonance with Empirical Mode Decomposition(EMD)de-noising under Levy noise is studied.Intrinsic Mode Functions(IMFs)decomposed from EMD under strong noise environment is difficult to recognize object signals.In order to solve the problem,a SR method under Levy noise after denoised by EMD decomposition is presented.Firstly,Levy noise is used as the complex background noise.Secondly,the noisy signals are decomposed by EMD,The first two layer IMFs with the largest SNR after decomposing are superimposed and averaged.Lastly,the average signal is processed with SR method to achieve the detection of the target signal.(2)Weak signal detection based on underdamped stochastic resonance with an exponential bistable potential is studied.The underdamped bistable SR(UBSR)potential structures are deficient to match with the complicated and diverse mechanical vibration signals and their parameters are selected subjectively which probably resulting in poor performance of UBSR.To overcome these shortcomings,this thesis proposes an underdamped SR with exponential potential(UESR)which is generalized by using a harmonic model and a Gaussian potential(GP)model.Firstly,the signal-to-noise ratio(SNR)formula is derived by studying UBSR potential function structure,and it proves the proposed system can generate a stochastic resonance phenomenon in theory.Then,the effects of system parameters on system performance are investigated by output SNR versus noise intensity D for different parameters.Finally,the proposed method is used to process bearing experimental data and further perform bearing fault diagnosis,and particle swarm optimization(PSO)algorithm is used to select the appropriate parameters.The experimental results demonstrate that a larger output SNR and higher spectrum peaks at fault characteristic frequencies can be obtained by the proposed method compared with the UBSR method,which confirm the effectiveness of the proposed method.(3)Weak signal recovery based on power function stochastic resonance is studied.Aiming at the problem that weak signals are difficult to recover under the strong noise environment,a power function recovery system is proposed.This thesis use crosscorrelation coefficient as the measurement index,and study the effect of different system parameters,noise intensity and signal amplitude on recovery performance.Firstly,the power function recovery equation is deduced based on the power function stochastic resonance equation,and the power function recovery system model is obtained.Then,the influence of different system parameters,noise intensity,and signal amplitude on the recovery performance is analyzed.Finally,power function recovery system achieves single-frequency sinusoidal signal,multi-frequency sinusoidal signal and single pulse signal recovery in the case of fewer sampling points and optimize parameters are optimized with the PSO algorithm.