Research On Stochastic Resonance Signal Recovery

In recent years, how to utilize noise and nonlinear systems to detect weak signals has become one of the research hotspots which have caused widespread concern. A lot of discussions were proposed, such as how to select an appropriate stochastic resonance model, how to achieve the best stochastic resonance state, how to extract and recover signals and so on. Among them, a further study needs to be done to reproduce accurately the original signal from the detected weak signal.In this paper, the restoring methods and techniques for noisy signals have been studied. Firstly, we analyzed in detail the influence of the three major factors in bi-stable systems (signals, noises and parameters of the systems) to the state of stochastic resonance and whereby ratiocinated the reason of waveform distortion in the bi-stable stochastic resonance process. Based on this, a recovering system was deduced. Next, according to the bi-stable system to configure the system parameters and then recovered the initial noisy periodic signal directly on both of the systems. Further analysis to the effect of the signals, noises and system parameters on the recovering process was done to optimize the recovery signal as well as to improve the accuracy, which result in two different self-adaptive parameter optimization methods.In order to calculate the output of stochastic resonance, the4th Runge-Kutta algorithm was used. Regarding the signal-to-noise ratio as an indicator to measure the state of stochastic resonance, we analyzed how various factors of bi-stable stochastic resonance systems affect the stochastic resonance, and then combined the signal processing with particle’s dynamics theory to revert signal’s trajectory, which led to the deduction of a recovery system. Next, based on the organic combination of these two systems, the initial noisy periodic signal was reached directly. According to the influence degree from various factors of the recovery system on the signal restoration as well as the relations between them, after sampling noisy periodic signal and estimating the noise variance and the signal frequency, the maximum signal-to-noise ratio was regarded as the target to optimize the parameters of the recovery system automatically. Alternatively, the optimized process was done automatically based on the deduced formulas to calculate the recovery system parameters.The simulations of the two self-adaptive recovery methods show that this method is effective and feasible. The simulations also find that the system can reproduce the noisy signal regardless that whether the signal can meet the requirements of stochastic resonance or not. This method has expanded the range of signal processing and will have a broad application prospects in actual projects.