Dynamic Bayesian wavelet transform as new methodology is proposed. Repetitive transients are extracted by using Dynamic Bayesian wavelet transform. The proposed method can be initialized by many fast detection algorithms. The proposed method is convergent as iterations increase. The proposed method can be extended to find the optimal parameters of other signal processing methods.