IGBT老化状态下基于BAS-SVM模型的结温预测方法

Junction Temperature Prediction Method Based on BAS-SVM Model in Aging State of IGBT

为进一步研究IGBT老化对其结温的影响,提出了一种新型的结温预测方法。通过功率循环加速老化实验,模拟模块实际运行老化状态,获取了不同老化程度下的饱和压降、集电极电流和结温数据,并进行分析。采用数据驱动的方法,建立了基于天牛须搜索算法优化支持向量机(BAS-SVM)的结温预测模型。结果表明,与粒子群算法-支持向量机(PSO-SVM)模型和天牛须搜索-BP神经网络(BAS-BP)模型相比,BAS-SVM模型更能有效缩短训练时间,收敛速度更快,且对IGBT结温的预测精度更高,是一种更有效的预测模型。

In order to further study the effect of IGBT aging on its junction temperature,a novel method for predicting junction temperature was proposed.The actual aging state of module was simulated by the power cycle accelerated aging experiment,and the saturation drop voltage,collector current and junction temperature data under different aging degrees were obtained for analyzed.By using the data-driven method,a junction temperature prediction model based on Beetle Antennae Search optimized Support Vector Machine(BAS-SVM)was established.The analysis results showed that compared with the Particle Swarm Optimization algorithm-Support Vector Machine(PSO-SVM)model and the Beetle Antennae Search-BP neural network(BAS-BP)model,the BAS-SVM model could shorten the training time more effectively,its convergence speed was faster,and the prediction accuracy of the IGBT junction temperature was higher,which made it a more effective prediction model.