IGBT Health Status Evaluation Model Based On Dynamic Bayesian Network

As environmental problems worsen,people pay more and more attention to the development and use of green energy.With its own advantages,IGBT modules are widely used in renewable energy power generation,new energy vehicles and other fields.Research data shows that IGBT module failure accounts for a very high proportion of all power device failures.Once the IGBT module fails,it will seriously affect the normal operation of the system,so it is necessary to monitor and evaluate the health of the IGBT module.When performing health status evaluation,the health status of the assessed object is generally polymorphic,and its evaluation should also be dynamic.In recent years,with the rise of Bayesian Networks,it has provided effective means for the research of multi-state systems.This paper proposes a Dynamic Bayesian Network-based IGBT module health evaluation model,which integrates multiple information to realize the dynamic evaluation of the IGBT module health status.The main research is as follows:First,after understanding the structural characteristics,failure mechanism and main failure modes of the IGBT module,perform FMEA analysis on the IGBT module,sort out the logical relationship among the failure modes,failure causes,and failure effects of the IGBT module,and provide information support for the next FMEA transfer to BN.Secondly,when IGBT module health assessment Dynamic Bayesian Network modeling,extract information from FMEA analysis,determine the node variables of the Bayesian Network and divide its state.The structure matrix is used to convert the information of the FMEA table into a Bayesian Network structure.When calculating network parameters,use fuzzy interval group decision-making method to solve the prior probability of root node that cannot be directly obtained.Use the NoisyMax model to find the conditional probability of the intermediate node with multiple states and multiple parents.Then according to the aging characteristics of the IGBT module,the transfer probability of the Dynamic Bayesian Network is determined,and the establishment of the Dynamic Bayesian Network is completed.According to the characteristics of Bayesian Network,a health assessment method suitable for Dynamic Bayesian Network is proposed.Finally,an example is applied to the Dynamic Bayesian Network model of IGBT module health assessment,and the application method of the model is introduced.According to the evaluation results of the module's health status and the failure reasoning results,overhaul and maintenance suggestions were put forward,which verified the feasibility and operability of the model.