For Weak Signal Recognition Based On Stochastic Resonance

Weak signal detection technology has been a public topic for the scholars. Rapid detection of weak characteristic signals under strong background noise to guarantee the safe operation of the large mechanical and electrical equipment and to avoid major accident happening in engineering projects has extensive application value.Any system must contain noise. When the measured signal is relative weak for the background noise which will make the measured signal has a low signal-to-noise ratio. Thus, it will make weak characteristic signal detection difficult.Stochastic resonance is different from the traditional signal detection methods (such as:wavelets transform method, Kalman filtering method, etc.). It does not need to filter out or suppress the noise energy and can transfer part of the noise energy into the characteristic signal. What is more, it can improve the system output signal-to-noise ratio and achieve the aim of identification of weak signal.For a given input signal and noise intensity, the effect of stochastic resonance of the system depends mainly on the system structure parameters. Therefore, design a kind of effectively method of adaptive system structure parameters which can get good effect of stochastic resonance. Usually structure parameters of the adaptive method based on signal-to-noise ratio as a measure. Furthermore, the SNR is the approximation which is deduced under the adiabatic approximation theory and has a certain error. If the input signal does not meet the condition of adiabatic approximation, the error of SNR expression will be big and even may be wrong.To avoid the disadvantages of signal-to-noise ratio, this paper proposes a new adaptive stochastic resonance method by adjusting the parameters of the structure of the bistable system. Considering stochastic resonance system characteristics and input/output signal characteristics, select the optimal structure parameters of the system. On this basis, further proposed based on wavelet transform method of adaptive stochastic resonance system, first through the stochastic resonance system, improve the energy of the signal; Then using wavelet threshold method to eliminate the noise in the signal, outline signal in the time domain waveform, identify the signal cycle.Through the analysis of the results of simulation experiment, found that the input signal for single frequency, dual frequency or multiple frequency periodic signals, this method can get excellent system structure parameters lead to obvious resonance phenomenon. What is more, it proves the validity and reliability of this method.