Research On Health Prediction Of Power Electronic Circuits Based On Model

The rapid development of science and technology has created a large number of large-scale industrial systems.The safe and stable operation of the system is inseparable from the support of fault diagnosis technology.At present,the model-based method is an important branch of fault diagnosis technology,and the difficulty lies in the establishment of the system model.Bond Graph(BG)is an energy-based modeling method that can effectively implement multi-energy domain hybrid system modeling.In recent years,domestic and foreign scholars have achieved a lot of practical results in the modeling of bond graphs.Taking into account the influence of various uncertainties in the actual system,this paper proposes to use the bond graph theory as a tool,citing linear fraction transformation(LFT),and combining interval algorithms to establish interval uncertainties.The system bond graph model,and then obtained Interval valued Analytical Redundancy Relations(I-ARRs),to achieve the hybrid system robust fault detection.The I-ARRs are very sensitive to the derivative of the measured signal and will affect the residual estimation.To solve this problem,a robust sliding mode differentiator is used to achieve an optimal estimation of the derivative of the noisy measurement signal,reduce the residual noise,reduce the false alarm and miss rate of the system,and improve the robustness of fault diagnosis.This method is applied to the Buck circuit and the experimental simulation results are analyzed to verify the effectiveness of the method.Taking into account the types of common faults in power electronic circuits,fault diagnosis and isolation of important components of the system are completed,and the remaining service life of the system is predicted.Using the interval resolution redundancy relationship,combined with the component degradation model,its corresponding residual degradation model is obtained.Using the intelligent prediction algorithm,the residual degradation mechanism data is taken as input,and the component degradation mechanism data is used as an output,and the global analytical redundancy relationship is inversely inferred.In combination with the failurerobustness threshold and the failure value of the components,the remaining useful life of components is predicted.