Study On Prediction And Control Strategy Of IGBT Module Junction Temperature

IGBT modules are used in many applications.However,IGBT modules often have some failure problems.There are many factors leading to the failure of IGBT modules.Among these factors,temperature affects IGBT modules.The impact of the module is very large.The large fluctuation of the junction temperature of the IGBT module will cause the device to be impacted by thermal stress for a long time,resulting in the fracture of the bonding wire,and thus the device failure,and long-term operation at high temperature and the temperature of the device exceeds its own temperature resistance value,which will accelerate The aging of the parts leads to failure.Therefore,from the perspective of improving the safety and reliability of the device,the thesis focuses on the IGBT module junction temperature prediction and junction temperature control methods.Based on the IGBT module electro-thermal coupling model,neural network and frequency conversion temperature control method,this thesis focuses on the two aspects of IGBT module junction temperature prediction and IGBT module junction temperature control.The main research contents include:(1)The electrothermal coupling model predicts the IGBT junction temperature.This thesis derives the average loss model of IGBT power devices under SPWM modulation,and establishes the thermal model of the power module.Through the establishment of the 3-D model of the IGBT module in COMSOL,the thermal characteristics of the thermal coefficient of the module material with temperature are taken into consideration.The transient thermal impedance curves of each RC unit of the thermal model are fitted,and the thermal parameters of the thermal model are obtained by fitting.The junction-case thermal resistance changes of the power device thermal model under different loss distributions and different IGBT chip loss ratios are studied,and the updated power module junction-case thermal resistance expression is obtained.the IGBT electrothermal coupling model based on simulink simulation and the IGBT electrothermal coupling model based on numerical iterative calculation are proposed.(2)The neural network predicts the junction temperature of the IGBT module.It is proposed to apply the extreme learning machine ELM to the junction temperature measurement of the IGBT module,and the ELM network is optimized by the particle swarm algorithm PSO.A constant temperature experiment platform for the power module was built,and data such as the collector current of the power module,the conduction voltage drop of the collector and emitter,and the junction temperature of the module were collected.The ELM network is established to predict the junction temperature of IGBT devices.Considering its own shortcomings will lead to unstable prediction and reduce the prediction accuracy,the PSO algorithm is used to optimize it.The prediction results show that the optimized effect and prediction accuracy have been furthe proved.(3)On the basis of the model presented in Chapter 3,the junction temperature of IGBT device is changed by changing the switching frequency.First,determine the range within which the switching frequency varies.the power module conduction current,module junction temperature and switching frequency function expressions are established,and the feasibility of the variable frequency temperature regulation strategy is verified through simulation.The experimental circuit with DSP28335 as the control board is built,and the drive circuit of the power device is designed.It is used to verify the control strategy proposed in this thesis.