Abstract:This paper discusses the health prediction of component level in dynamic system. The residual service life of the component level is obtained by means of inverse residual signal. Firstly, the dynamic model of the system is established by using the bond graph, and the global analytical redundancy is obtained, and the fault detection and isolation is carried out by fault feature matrix. Then, the residual degradation curve corresponding to the components is obtained by combining the degradation mechanism of the components, so as to obtain the data samples of the degradation process of the components. According to the relationship between the equivalent resistance and residual degradation of components, the remaining use time of component level can be calculated by combining the fault threshold with the failure value and the Extreme Learning Machine (ELM) algorithm. Finally, this method is applied to the typical power electronic circuit Buck circuit, and the simulation is carried out in the 20-sim software and Matlab simulation environment to verify the effectiveness of this method.