The selection of the parameters of the stochastic resonance system plays a decisive role in the effect of the output signal. In this paper

aiming at the problem that the stochastic resonance cannot process many kinds of weak communication signals in general

an enhancement method based on adaptive parameter-tuning stochastic resonance is proposed. Firstly

the energy transfer nature of the stochastic resonance is explained from the characteristic subspace of the signal

and a measuring function based on singular value decomposition is presented as the evaluation function. Secondly

after analyzing the effects of two different system parameters

the amplitude normalization is utilized to optimize one single parameter

which reduces the complexity. And the sliding average filter is added to the module to prevent the phenomenon of the amplitude drift. Finally

based on the artificial fish swarm algorithm

the overall framework and specific steps of the method are modularly designed. The simulation result demonstrates that the proposed method can match the noisy signal and the nonlinear system with an average of 4 to 5 iteration convergence speeds for four kinds of signals.