Modeling Research On IGBT Module Junction Temperature Prediction Model

Insulated Gate Bipolar Transistor(IGBT)is the core component of energy conversion and transmission,and is widely used in smart grid,rail transit,aerospace and new energy equipment.However,the fluctuation of the IGBT junction temperature will cause the module to age,reduce the working efficiency of the power converter device,and affect the safe and reliable operation of the system.Therefore,the collector current,collector-emitter voltage,junction-to-case thermal resistance and junction temperature of IGBT module were measured by power cycle accelerated aging experiment,single pulse experiment and junction-to-case thermal resistance measurement experiment,and based on the measured experimental data,the IGBT junction temperature prediction models were established.The specific research contents are as follows:(1)The working principle and working characteristics of the IGBT module were expounded in detail.And the failure mechanism of the IGBT module was analyzed from the aspects of package failure and chip failure.But the bonding wires dropout and solder layer fatigue in package failure were the main failure modes of IGBT module,which provides a theoretical guidance for studying the relationship between IGBT junction temperature and the collector-emitter voltage,and the relationship between IGBT junction temperature and junction-to-case thermal resistance.(2)The structure of the IGBT module used in this paper was introduced in detail.In order to make the IGBT module more in line with the actual working conditions,the aging of the module was accelerated by the?T_cpower cycle accelerated aging experiment.In the power cycle process,the collector current,collector-emitter voltage,and junction temperature of the IGBT were measured by the single pulse experiment,and the junction-to-case thermal resistance of the IGBT module was measured by the junction-to-case thermal resistance measurement experiment.Based on the experimental data,the relationships between the above parameters were studied,and a new IGBT junction temperature calculation method considering aging of IGBT module was obtained.The method can be used in applications where the working environment is not demanding,however,in the harsh working environment,it is necessary to further study the more accurate IGBT junction temperature prediction model.(3)Based on the basic principles of support vector machine,particle swarm optimization algorithm and chicken swarm optimization algorithm,the collector current,collector-emitter voltage and junction-to-case thermal resistance of the IGBT module as input,and the IGBT junction temperature as output,the junction temperature prediction model of particle swarm optimization support vector machine(PSO-SVM)and chicken swarm optimization support vector machine(CSO-SVM)were established.The ratio of the average absolute error of the PSO-SVM and CSO-SVM prediction models is 1:0.5988,the ratio of the average relative error is 1:0.6097,and the ratio of the root mean square error is1:0.5170.Therefore,CSO-SVM prediction model has better prediction performance than PSO-SVM prediction model,but there is still room for improvement in prediction accuracy and convergence speed.(4)In order to improve the prediction accuracy and convergence speed of IGBT junction temperature prediction model,an improved chicken swarm optimization support vector machine(ICSO-SVM)junction temperature prediction model was proposed in this paper.By comparing the results of the three prediction models,the ratio of the average absolute error of the PSO-SVM,CSO-SVM and ICSO-SVM prediction models is1:0.5988:0.4739,the ratio of the average relative error is 1:0.6097:0.3909,and the ratio of the root mean square error is 1:0.5170:0.4487.Therefore,ICSO-SVM prediction model has better prediction performance than PSO-SVM and CSO-SVM prediction models,which provides a theoretical basis for the reliable operation of IGBT module.