基于改进GSO-SRU的整流二极管寿命预测  

Life prediction of rectifier diodes based on improved GSO-SRU

作　　者：王亚东 杨磊 WANG Yadong;YANG Lei(CNNC Operation and Maintenance Technology Co.,Ltd.,Hangzhou 311215,China)

机构地区：[1]中核运维技术有限公司,杭州311215

出　　处：《电子测试》2023年第5期40-46,共7页Electronic Test

摘　　要：变频调速装置作为节能减耗的关键设备,在生产线上得到广泛的应用。针对其中功率模块的寿命进行有效预测,有助于提高系统可靠性。本文提出一种基于改进萤火虫优化算法(GSO)和简单循环单元(SRU)神经网络的整流二极管剩余寿命预测模型。首先通过加速老化试验得出二极管的结温和导通压降数据,将数据集进行划分后将训练集中的结温和导通压降作为特征参数,进行数据预处理后输入到经改进GSO算法优化超参数的SRU网络模型中完成训练,实现对整流二极管剩余寿命的预测。最后通过测试集数据验证,并与不同预测模型结果作对比,证明该预测模型具有十分优良的准确性,提高了寿命预测精度,为功率模块安全稳定运行提供了保障。As a key equipment for energy conservation and consumption reduction,frequency conversion speed regulation devices are widely used in production lines.Effectively predicting the lifespan of the power module can help improve system reliability.This article proposes a residual life prediction model for rectifier diodes based on an improved glowworm swarm optimization(GSO)and a simple recurrent unit(SRU)neural network.Firstly,the junction temperature and conduction voltage drop data of the diode were obtained through accelerated aging tests.After dividing the dataset,the junction temperature and conduction voltage drop in the training set were used as characteristic parameters.After data preprocessing,they were input into the SRU network model optimized by the improved GSO algorithm to complete the training and achieve the prediction of the remaining life of the rectifier diode.Finally,through the validation of the test set data and comparison with the results of different prediction models,it was proven that the prediction model has excellent accuracy,improves the accuracy of life prediction,and provides a guarantee for the safe and stable operation of the power module.

关 键 词：高压变频器 整流二极管 加速老化试验 寿命评估 循环神经网络 

分 类 号：TN710[电子电信—电路与系统]