| IGBT,as a new generation of power semiconductor devices,is developing in the direction of high frequency,high power and high integration,so the related problems such as electricity and heat are more and more prominent.IGBT may produce a large amount of heat dissipation and large temperature rise,this has a great influence on device reliability.The switching loss is a portion of device loss,but how to access to accurate switching loss value simple and fast to be solved.Therefore,there have an important significance on researching on dynamic characteristics and switch loss of IGBT module to improve the device performance and reliability.Paper's main work is as follows:In this paper,we used IGBT power module dynamic testing system for dynamic test and obtained corresponding dynamic waveform in order to study the power module IGBT switching loss.This laid a foundation for the IGBT switching loss modeling.Through the switch dynamic waveform to obtain accurate switching loss value.The paper analyzed the trend of switching loss as the working frequency and collector current changes.Then established simple model of the switching loss.In this paper,we adopted the artificial intelligence model(neural network,support vector machine)to build IGBT module loss model.This paper also determined the parameters of the model to establish the BP neural network model of switching loss of IGBT.This switching loss model is based on large data.Besides,This paper also builded a switching loss model with the algorithm of SVM.The precision and the effect of this model is both better compared with neural network model.This made up for the power module IGBT switching loss defects in the modeling of small sample.In this paper,we conducted aging test of IGBT by aging test platform and obtained power module switching loss values of different degree of aging,Then this paper will establish switching loss aging model through analyzing the aging of the measured power module switching loss of the data. |
PDF Full Text Request