Weak Impact Signal Detection Technology Study Based On Stochastic Resonance

Weak signal detection is a traditional investigative task based on strong noise background, which has extensive applications in engineering field. For example, the locating of weak signal source plays an important role in fault diagnosis and post-disaster rescue. Weak signal detection technology which is based on electronics, information theory, physics, statistics, etc., is to study statistical characteristics and correlation of the measured signals and noise. Then it will exert a series of signal processing methods to control noise and detect the weak signals in strong background noise finally.It is usually believed that the noise is not conductive to signal detection. It weakens the system output response to the signal, making the signal submerged in a noisy background noise difficult to be detected. However, recently incessant development of nonlinear science, especially the proposed stochastic resonance theory, becomes a novel idea for weak signal detection. It pointed out that by adding certain amount of noise in order to enhance the output capacity of nonlinear systems, thus enhancing the detection of weak signals. Focusing on non-periodic impact signal, this thesis presents the principle and practicable method to utilize stochastic resonance for weak signal detection. The main research and gains of this thesis are organized as follows:Firstly, this paper reviews theories related to weak signal detection, especially the development of stochastic resonance and experimental study. By using the two representational non-linear equations, which called Langevin equation and Fokker-Planck equation, this thesis has introduced the basic principles of stochastic resonance, and studied the relationships among bi-stable stochastic resonance system, signal and noise by a mount of simulations.Secondly, focusing on weak non-periodic impact signal, this thesis studied the impact of the structural parameters of the bi-stable systems, noise and reference signal characteristic parameters on signal detection performance. The stochastic resonance effect of impact signal and colored noise through bi-stable stochastic resonance system has been studied by a great of simulations and experiments.Thirdly, for extracting weak impact signals from strong colored noise background, detailed design ideas and processes of weak signal extraction algorithm based on adaptive stochastic resonance has been discussed. This algorithm has been realized mainly by adaptively adjusting the system parameters or adding input signal gains to enhance stochastic resonance effect, so as to extract the signal under test.At last, according to the arithmetic designed above, a weak signal detection simulation system based on stochastic resonance, which is written by MATLAB scripts, is realized and applied to a practical recording sound file.