In dynamic fault tree analysis method, the model of complex system is constructed by fault tree, and the solution process is implemented by Markov process. It is provided with both the advantages of fault tree analysis and Markov process, which makes it extensively used in many engineering fields. Bayesian networks have the powerful capacity of dealing with representation, quantification and inference for uncertain information. Rectifier feedback system is an important part of electrical control systems of large mining excavator, whose working reliability has a significant impact on the reliability of the entire system. There exist a large number of redundant structures in the is system. The continuous time Bayesian network based modelling and analysis method is applied to model and evaluate these dynamic characteristics. Furthermore, the sample size of these complex systems is poor, which makes the precise estimation of failure parameters very difficult. The fuzzy numbers are utilized to characterize the failure rate of components. The proposed method is carried out on rectifier feedback system to demonstrate its effectiveness.