中文摘要：

功率半导体器件的开关损耗的准确预测和评估对研究系统设计、选择合适的散热系统和提高系统的可靠性都是很重要的。采用功率半导体器件的开关动态测试系统,可以自动调整直流母线电压、集电极电流、门极驱动电压和开关频率,记录开关动态电压、电流波形,计算获得大量的损耗数据。通过建立人工神经网络模型对开关损耗预测,并改变模型参数进行分析与研究,获取最佳模型。

英文摘要：

The accurate prediction and evaluation for switching loss of power semiconductor devices are very important to study the system design, select the suitable cooling system, and improve the system reliability. The switch dynamic testing system of power semiconductor devices can automatically adjust the DC bus voltage, collector current, gate driving voltage and switching frequency, record the switch dynamic voltage and current waveforms, and obtain the massive loss data by calculation. The neural network model was established to predict the switching loss, and change the model parameters for analysis and study, so as to obtain the best model.