基于GA-BP算法的IGBT结温预测模型
|
摘要
人们已经使用BP神经网络进行IGBT结温预测,但BP神经网络结温预测模型却存在着预测误差较大的问题。该文基于英飞凌IGBT模块,选取了饱和导通压降和集电极电流的温敏参数来预测器件结温,分别测量了集电极电流大于和小于临界电流2组实验数据,建立了基于GA-BP算法和BP神经网络的结温预测模型。实验结果表明:当集电极电流大于临界电流时,两种算法预测精度更高;使用GA-BP算法结温预测模型较BP神经网络结温预测模型更接近实测数据,预测误差更低。
BP neural network has been used to predict IGBT junction temperature,but the BP neural network junction temperature prediction model has a large prediction error. Based on Infineon IGBT module,the temperature sensitivity parameters of saturation conduction voltage drop and collector current are selected to predict the junction temperature of the device. The experimental data of collector current greater than and less than the critical current are measured respectively. The junction temperature prediction model of the network and GA-BP algorithm. The experimental results show that the prediction accuracy of the two algorithms is higher when the collector current is greater than the critical current. The junction temperature prediction model using GA-BP algorithm is closer to the measured data than the BP neural network junction temperature prediction model,and the prediction error is lower.
BP neural network has been used to predict IGBT junction temperature,but the BP neural network junction temperature prediction model has a large prediction error. Based on Infineon IGBT module,the temperature sensitivity parameters of saturation conduction voltage drop and collector current are selected to predict the junction temperature of the device. The experimental data of collector current greater than and less than the critical current are measured respectively. The junction temperature prediction model of the network and GA-BP algorithm. The experimental results show that the prediction accuracy of the two algorithms is higher when the collector current is greater than the critical current. The junction temperature prediction model using GA-BP algorithm is closer to the measured data than the BP neural network junction temperature prediction model,and the prediction error is lower.
引文
[1]钱照明,张军明,盛况.电力电子器件及其应用的现状和发展[J].中国电机工程学报,2014,34(29):5149-5161.
[2]杨旭,周雒维,杜雄,等.绝缘栅双极型晶体管结温测量方法及其发展[J].电测与仪表,2012,49(2):7-12.
[3]姚芳,王少杰,李志刚.基于IGBT模块饱和压降温度特性的结温探测研究[J].电子器件,2017,40(2):291-295.
[4]姚芳,胡洋,李铮,等.IGBT模块结温预测BP神经网络模型[J].西北大学学报:自然科学版,2017,47(3):367-374.
[5]刘春艳,凌建春,寇林元,等. GA-BP神经网络与BP神经网络性能比较[J].中国卫生统计,2013,30(2):173-176.
[6]姚仲敏,潘飞,沈玉会,等.基于GA-BP和POS-BP神经网络的光伏电站出力短期预测[J].电力系统保护与控制,2015,43(20):83-89.
[7]王志斌,孔亚楠,刘永成,等.遗传算法优化BP神经网络的大功率LED结温预测[J].光电子.激光,2014,25(7):1303-1309.
[8]彭基伟,吕文华,行鸿彦,等.基于改进GA-BP神经网络的湿度传感器的温度补偿[J].仪器仪表学报,2013,34(1):153-160.
[9]吴志杰,孔凡敏,李康.基于遗传算法的BP神经网络的LED寿命预测模型[J].半导体技术,2018,43(5):375-380.
[10]周志宇.基于EMD与GA-BP神经网络的短期负荷预测[J].电测与仪表,2013,50(4):17-21.
[2]杨旭,周雒维,杜雄,等.绝缘栅双极型晶体管结温测量方法及其发展[J].电测与仪表,2012,49(2):7-12.
[3]姚芳,王少杰,李志刚.基于IGBT模块饱和压降温度特性的结温探测研究[J].电子器件,2017,40(2):291-295.
[4]姚芳,胡洋,李铮,等.IGBT模块结温预测BP神经网络模型[J].西北大学学报:自然科学版,2017,47(3):367-374.
[5]刘春艳,凌建春,寇林元,等. GA-BP神经网络与BP神经网络性能比较[J].中国卫生统计,2013,30(2):173-176.
[6]姚仲敏,潘飞,沈玉会,等.基于GA-BP和POS-BP神经网络的光伏电站出力短期预测[J].电力系统保护与控制,2015,43(20):83-89.
[7]王志斌,孔亚楠,刘永成,等.遗传算法优化BP神经网络的大功率LED结温预测[J].光电子.激光,2014,25(7):1303-1309.
[8]彭基伟,吕文华,行鸿彦,等.基于改进GA-BP神经网络的湿度传感器的温度补偿[J].仪器仪表学报,2013,34(1):153-160.
[9]吴志杰,孔凡敏,李康.基于遗传算法的BP神经网络的LED寿命预测模型[J].半导体技术,2018,43(5):375-380.
[10]周志宇.基于EMD与GA-BP神经网络的短期负荷预测[J].电测与仪表,2013,50(4):17-21.