Weak Signal Detection Technology Study Based On Stochastic Resonance

An important aspect in information science field is to acquire real information of objective world, but there exists noise in all systems, especially it is difficult to detect weak signal in strong noise. More and more attention has been paid to the problem of how to measure the weak signal in strong noise. And, now, detecting weak signals from strong noise background is valuable in engineering field. Some characteristic frequency signal detection to avoid the accident played an important role. Weak signal detection technology which is based on electronics, information theory, physics, statistics, etc., is to analyze principles and rules of the noise and study statistical characteristics and correlation of the measured signals and noise. Then it exerts a series of signal processing methods to control noise and detect the weak signals in strong background noise.By reading, experiments, comparing existent detection methods at home and abroad, and taking advantage of the new technologies brought about by computer science in weak signal detection, the author systematically studied weak signal detection theories and extraction methods. This dissertation focused on non-linear dynamical principles in signal detection, especially stochastic resonance. On the basis of studying stochastic resonance theory(SR), used a lot of simulations to analyze and verify the basic theory of SR, and aiming at explorer and study actual signals which are multi-frequency weak signal and colored noise in engineering tests.This dissertation also designed the adaptive weak signal extraction arithmetic based on stochastic resonance. Through adjust adaptively the parameters of the system to state of SR, the system extract the multi-frequency weak signal.In order to meet the needs of engineering, a convenient and practical weak signal detection system has been developed by using MATLAB, C/ C++, COM, MATCOM. As a separate system, the system is not to need to run in MATLAB, filling in the blank of previous system only run in the MATLAB. Finally, the verification from a recording file has testified its validity.