Research On Smart Fault Prediction Of The Power Electronics Based On Models

Power electronic technology plays very important roles in the aerospace, wind power and otherfields, and which could efficiently achieve energy transformation. However, power electronic circuitsusually work in high frequency switch working way, which could pose increased challenges for thestability of equipment and the safety of workers. Once the power electronic circuits occur fault, itwill cause very serious consequences. Therefore, the application of intelligent algorithms to the faultsprediction of power electronic circuits can effectively warn systems failure, It can achieveforeseen maintenance and preventive maintenance instead of traditional maintenance. So the researchon the failure prediction technology of power electronic circuits is essential.This thesis mainly focuses on the research of fault simulation and prediction of the powerelectronic circuits. The research content contains following aspects:（1）The analysis on the mainly component equivalent circuit modeling and its failuremechanism. Through analyzing the equivalent circuit modeling and its failure mechanism of powerMOSFET, electrolytic capacitors, diodes, and the inductance of the equivalent circuit model and thefailure mechanism to determine the failure parameters.（2）Establishing the bond graph model of typical power electronic circuits and implementingsimulation with fault injection. The application of bond graph theory in circuit system is studied, theideal and non-ideal bond graph model of the Buck converter and Boost converter are set up, andobserve the fault simulation waveform with on-line fault injection. Which verifing the theory that theincreases of equivalent series resistance of electrolytic capacitor will causes the increase of the outputvoltage ripple. According to the average model of no-ideal Buck converter and no-ideal Boostconverter, the average bond graph model of them is established, which can be used to observe thesystems’output more efficiently.（3）The fault prediction method of power electronic circuits about components-level andsystem-level are studied. The prediction algorithm of particle filter、Kalman Filter and the GreySystem is studied.