摘要
针对绝缘栅双极型晶体管(IGBT)老化特征预测精度不高的问题,提出了改进鲸鱼算法(IWOA)与BP神经网络结合的IGBT老化特征预测方法。首先,分析IGBT失效模式,选取集射极关断尖峰电压(Vce-p)和IGBT模块内部的热敏电阻阻值(Rntc)作为监测指标。接着,搭建基于BP神经网络的IGBT老化状态预测模型,针对BP神经网络训练效率低、精度不足的问题,采用鲸鱼算法对BP神经网络输入权值和隐含层阈值进行优化。然后,针对鲸鱼算法全局优化能力弱和易陷入局部最优等问题,进一步融合随机性学习策略和柯西变异策略进行改进。最后,搭建仿真环境完成对基于改进鲸鱼算法优化BP神经网络(IWOA-BP)的IGBT老化特征预测模型的建模,并使用获取的老化数据分别对所提预测模型进行训练和验证,再与不同预测模型作对比。仿真结果显示,IWOA-BP预测模型在IGBT老化预测中表现出更高的精度和适用性。
To address the problem of low prediction accuracy of Insulated Gate Bipolar Transistor(IGBT)aging characteristics,a prediction method for IGBT aging characteristics combining the Improved Whale Optimization Algorithm(IWOA)and Back Propagation(BP)neural network is proposed.Firstly,the IGBT failure modes are analyzed.The collector-emitter turn-off spike voltage(Vce-p)and the resistance value of the thermistor(Rntc)inside the IGBT module are selected as monitoring indicators.Secondly,an IGBT aging state prediction model based on the BP neural network is constructed.Aiming at the problems of low training efficiency and insufficient accuracy of the BP neural network,the whale optimization algorithm is adopted to optimize the input weights and hidden layer thresholds of the BP neural network.However,in view of the weak global optimization ability of the whale optimization algorithm and its tendency to fall into local optima,a random learning strategy and Cauchy mutation are further integrated for improvement.Finally,a simulation environment is set up to complete the modeling of the IGBT aging characteristic prediction model based on the improved whale optimization algorithm optimized BP neural network(IWOA-BP).The obtained aging data are used to train and validate the proposed IWOA-BP based IGBT aging characteristic prediction model,and comparisons are made with different prediction models.The simulation results show that the IWOA-BP prediction model exhibits higher accuracy and applicability in IGBT aging prediction.
作者
邱枫
吴棋龙
陈军
胡海林
QIU Feng;WU Qi-long;CHEN Jun;HU Hai-lin(School of Electrical Engineering and Automation,Jiangxi University of Science and Technology,Ganzhou 341000,China;Hangzhou Liwode Power Supply Co,Ltd,Hangzhou 310000,China)
出处
《磁性材料及器件》
2025年第3期94-101,共8页
Journal of Magnetic Materials and Devices
基金
国家自然科学基金项目(52262050)。
关键词
绝缘栅双极型晶体管
IGBT老化特征预测
BP神经网络
改进鲸鱼算法
IWOA-BP预测模型
Insulated Gate Bipolar Transistor
IGBT Aging Characteristics Prediction
Back Propagation Neural Network
Improved Whale Optimization Algorithm
IWOA-BP Prediction Model
