Abstract: The performance of stochastic resonance methods is mostly decided by its system parameters. The existing stochastic resonance methods have the fatal problems; for example, subjectively selecting parameters or optimizing only one parameter therefore ignoring the interactive effect between parameters. To solve the problems mentioned above, a new adaptive stochastic resonance method is proposed. Compared with the existing methods, the proposed method utilizes the optimization ability of ant colony algorithms, synchronously selecting and optimizing multiple system parameters and considering the interactive effect between parameters, and adaptively realizes the optimal stochastic resonance system matching input signals. Thus, the problems in selecting parameters are solved by using the proposed method. Therefore noise is weakened and weak characteristics are enhanced effectively, and the early faults are diagnosed accurately as well. Both simulations and a real case of locomotive rolling element bearings with an early fault demonstrate that the proposed adaptive stochastic resonance method obtains the improved results compared with the existing methods.

Key words: Adaptive stochastic resonance, Fault diagnosis, Multiple parameter optimization, Weak feature extraction