The stochastic resonance algorithm is widely used to process weak noisy signals with a single frequency. In practice, most signals are co-existing with multiple target frequencies, so stochastic resonance has certain limitations on processing multi-frequency signals. Adaptive adjustment of stochastic resonance parameters according to the target signal frequency is the key point to efficiently process weak multi-frequency signals. The simulation data illustrate that the Grey wolf algorithm can be used to optimize the structural parameters of the bi-stable stochastic resonance system. Aiming at the difference of multi-frequency signals, the Grey wolf algorithm can grope the best function parameters through realizing algorithm iteration and realize the efficient processing of multi-frequency weak signals.