Detection Of Weak Signals Under Strong Noise Background

Aiming at the problem that the spike impulsive noise in some channel environments no longer obeys the Gauss distribution law,and the weak mechanical fault signal in complex noise is difficult to identify,the detection of multiple types of signals in coupled monostable stochastic resonance systems is studied under the background of Levy noise.Firstly,a new measure criterion(kurtosis-interconnection index,referred to as KI)is constructed as the detection index of impulse signal on the basis of kurtosis index.Secondly,the construction principle of the coupled system is analyzed.The phase diagram and density function distribution of micro-particles are theoretically analyzed in periodic potential system under different noise intensities,and the mechanism of stochastic resonance is explained from the point of view of micro-particle motion.Then the method of parameter adjustment is described in detail,and the law of system output changing with parameters is analyzed.Finally,the system is applied to the application of mechanical fault signal.The analysis of a large number of experimental data shows that the input signal of the system can promote the transition of micro-particles in the system to produce stochastic resonance,and the variation of the system output x(t)with time t satisfies the theoretical analysis of the attractor curve.Compared with the kurtosis index as the detection tool of impulse signal,KI can increase the characteristic amplitude of the system output signal more than twice,and KI can be used as an effective detection index of impulse signal.This detection index can take into account the spike characteristics of the impact signal and the original characteristics of the impact signal,make up for the shortcomings of the kurtosis index as the detection tool of the impact signal,and enhance the anti-jamming performance of the system.The parameter interval of the system does not change with the change of noise characteristic parameter alpha when the output signal-to-noise ratio of the system is optimal,and the effect of stochastic resonance induced by coupled monostable system is better than that of monostable system.The engineering example data proves that this method can extract the characteristic information of fault signal effectively and can be applied to the detection of actual mechanical fault signal.