Stochastic resonance is an important technique applied in weak signal detection. The timedelayed feedback ecological vegetation growth (EVG) system driven by weak periodic signals and additive white Gaussian noise is used as a model to carry out detailed stochastic resonance phenomenon analysis, and is applied to weak signal detection and bearing fault diagnosis in this paper. Firstly, the FokkerPlanck equation is used to deduce the equivalent potential function and further obtain the expression of the system signaltonoise ratio. Then the correlation curve diagrams are used to concretely analyze the influence of different system parameters on potential function and signaltonoise ratio. The study results show that adjusting the system parameters, signal amplitude and noise intensity all can induce the timedelayed EVG system to generate the stochastic resonance phenomenon. Finally, through adjusting the parameters and using the stochastic resonance method of timedelayed feedback EVG system, it is successfully detected that the target frequency of the weak signal is 001 Hz and its amplitude is 2 978; and the obvious peak values are detected at the fault characteristic frequencies of the inner and outer rings of the bearing.