中高压IGBT开关特性的遗传神经网络预测  

Genetic Neural Network Prediction on Medium and High Voltage IGBT Switching Performance

作　　者：陈娜[1] 李鹏[1] 江剑[1] 邓焰[1] 何湘宁[1] 

机构地区：[1]浙江大学电力电子技术国家专业实验室,杭州310027

出　　处：《电工技术学报》2013年第2期239-247,254,共10页Transactions of China Electrotechnical Society

基　　金：国家自然科学基金(50737002);国家科技重大专项(2011ZX02604001)资助项目

摘　　要：中高压绝缘栅双极型晶体管(IGBT)的开关特性在电力变换器设计、变换器性能、效率和寿命改善中至关重要。本文基于中高压功率模块离线测试平台的数据,分析了工作环境如门极电压、门极电阻、集电极电压、工作电流和器件结温等参数对IGBT在感性负载电路中开关特性的影响,对开关特性各参数建立了基于遗传算法优化的误差反向传播多层前馈神经网络模型,实现了在额定值范围内对各种环境条件下的IGBT开关特性参数如开关时间、开关损耗、最大电流尖峰和最大电压尖峰的可靠预测。The switching performance of medium and high voltage insulated gate bipolar transistor （IGBT） is important in converter design and converter performance, efficiency and lifetime improvement. Based on the experimental data of an off-line medium and high voltage power module test bench, the influence of environmental parameter such as gate voltage and gate resistor, collector-emitter voltage, collector current and device junction temperature on device switching characteristics was explored and an error back-propagation multi-layer feed-forward neural network model based on genetic algorithm optimization was built in this paper. The model has realized reliable predictions of IGBT switching characteristics such as switching time, switching losses, voltage overshoot and current spike under different environments with high precision.

关 键 词：IGBT 开关特性 遗传算法 神经网络预测 

分 类 号：TN323.4[电子电信—物理电子学] TN386.2