Research On Health Condition Monitoring Technology Of IGBT Power Module Bond Wires

As the core device of electrical energy conversion,power electronic converters have been widely used in new energy generation,transportation,aerospace and other fields.Insulated Gate Bipolar Transistor(IGBT)is the most important component of power electronic converters and the most widely used power device at present.Its reliability greatly affects the reliability of the converter system.Therefore,how to evaluate its health condition has attracted the attention of many researchers.However,the bond wires which are responsible for the connection between chip and electrical terminal in IGBT module is relatively fragile,and the failure caused by their breaking or falling off is the main failure type of IGBT module.Therefore,aiming at the problem of bond wires aging failure,this paper focuses on the research work on failure analysis and health condition evaluation of bond wires.Firstly,the physical structure,working principle and characteristics of IGBT are introduced,and the failure types and failure mechanisms of IGBT module are summarized and outlined: chip-level failure and package-level failure.The effect of aging failure of the bonding wires is analyzed,that is,aging failure of the bond wires leads to an increase in their equivalent resistance,which in turn causes an increase in the saturation voltage drop of the IGBT module under the same conditions,providing a theoretical basis for subsequent research.Secondly,considering that the electric,thermal and stress fields exist inside the module at the same time and are coupled with each other when the IGBT module is operated under actual working conditions,a finite element simulation model of the IGBT module based on multi-physics field coupling is established.Simulation tests are performed under steady-state heat transfer and power cycle conditions respectively,and the temperature and stress distribution results are analyzed.In view of the influence of bond wires failure on the reliability of IGBT module,the IGBT module models with different number of bond wire shedding are established and simulated.The change law of temperature and stress distribution are analyzed,and the evolution law of bond wires failure is summarized.The results of the analysis show that: first,the bond wire in the middle is subjected to the highest stress and is the first to fail;second,with the increase of the number of bond wire off,the temperature fluctuation and the stress of the remaining bond wires becomes larger,and the speed of failure is accelerated;third,the failure evolution of bond wires is a positive feedback process.Finally,the parameters,such as collector current,temperature and saturation voltage drop,which characterize the health condition of bond wires are analyzed and selected.An experimental platform is designed and built to validate the selection of characterization parameters,and the data sets of characterization parameters are collected for different bond wires health condition.According to the relevant theory of intelligent algorithm,a model for evaluating the health condition of bond wires based on the improved support vector machine with genetic gray wolf optimization algorithm is studied.The model is trained and tested with the actual measured data sets,and the average evaluation accuracy is 93.26%,which can effectively realize the evaluation of the health condition of bond wires.And the average evaluation accuracy is improved by 2.55% compared to the improved evaluation model based on the genetic algorithm.The average convergence speed of the model training is improved by about 40% compared to the improved evaluation model based on the Gray Wolf optimization algorithm.From the two aspects of evaluation accuracy and convergence speed,the evaluation model studied in this paper has more advantages.