The model of bistable stochastic resonance system has a very good application property in weak signal detection, provides a new way of thinking to detect weak signals. It is a hot topic in recent years that looking for a more simple way to break through adiabatic approximation condition, so that it applies a high frequency signal detection. This paper studies the variation of the output characteristics of the system from bistable state ( ) to instability ( ) and then single stable state ( or ) , depending on the affect of the shape characteristics of the potential function. This paper presents one kind of parameter adjustment method based on potential function, by controlling the system parameters reduces adjustable parameter to one, without changing the system’s optimum output SNR, will reduce the workload of parameter adjustment. The simulation and experimental data collection prove that the method is simple, and has a certain value.