基于优化长短期记忆神经网络的IGBT寿命预测模型

doi:10.19595/j.cnki.1000-6753.tces.222231

摘要
图/表
参考文献
相关文章 (15)

全文: PDF (903 KB)
输出: BibTeX | EndNote (RIS)

摘要

为了防止绝缘栅双极性晶体管（IGBT）突发性失效影响电力电子设备安全可靠运行,急需对IGBT剩余寿命作出精确预测,这对现有预测模型在高准确性和低不确定性方面提出了挑战。该文提出一种优化模型,该模型通过利用逐次变分模态分解（SVMD）技术来提取退化特征,并采用贝叶斯方法优化长短期记忆（LSTM）神经网络的超参数以提高预测性能。首先,该模型通过SVMD技术将退化特征数据分解为多个模态后将有用模态重构从而提取和增强退化特征。其次,本文利用贝叶斯优化方法通过高斯过程（GP）代理模型和期望改进（EI）采集函数对LSTM预测模型超参数实现全局寻优。最后,基于SVMD特征提取技术和贝叶斯优化LSTM网络的预测模型通过实际IGBT退化特征数据证明了模型的有效性和优越性。结果表明：该文模型与传统优化模型相比,提高了13%的寿命预测准确性,并减少了34%的预测不确定性。

	服务

	把本文推荐给朋友
	加入我的书架
	加入引用管理器
	E-mail Alert
	RSS
	作者相关文章
	任宏宇
	余瑶怡
	杜雄
	刘俊良
	周君洁

关键词 ： IGBT, 可靠性, 寿命预测, 模态分解, 失效分布

Abstract：

Insulated Gate Bipolar Transistors (IGBTs) are the core components of power electronic systems for the conversion and control of electrical energy. However, the reliability of IGBT is not as high as expected due to the complex environment and operating conditions, and the sudden failure of IGBT will lead to unplanned downtime of the entire system. Therefore, the assessment of the remaining useful lifetime (RUL) of IGBT will help to guide the regular maintenance and reduce economic losses. To prevent the sudden failure of IGBT, it is urgent to accurately predict the RUL of IGBT, but most existing methods suffer from low prediction accuracy and high uncertainty. To address this problem, this paper proposes an IGBT life prediction model based on optimized long short-term memory (LSTM). Starting from the two cores of the data-driven model, optimize and upgrade “data” and “model”, which can effectively improve the accuracy and reduce the uncertainty of the model prediction.
Firstly, the raw condition monitoring (CM) data often contains many contaminated data that appears abnormal due to environmental interference and limitations of measurement technology. Meanwhile, CM data may also appear abnormal when IGBT devices degrade or fail. This data contains important information to characterize degradation and failure of IGBT. It cannot be processed simultaneously with contaminated data. The proposed model extracts and enhances degraded features by decomposing the IGBT degraded data into multiple modes using the successive variational mode decomposition (SVMD) technique and then reconstructing the useful modes. Secondly, the selection of the model’s hyperparameters will greatly affect the learning ability and training effect of the model. Traditionally, the selection of hyperparameters by empirical trial-and-error method has contingency and randomness, which will seriously affect the performance of the model. The proposed model uses Bayesian optimization (BO) method to realize global optimization of multiple hyperparameters in the model through Gaussian process (GP) proxy model and expectation improvement (EI) acquisition function. Finally, the effectiveness and superiority of the LSTM prediction model based on SVMD and BO is verified with real IGBT degradation data.
The validation results show that the predicted RUL is not very close to the real RUL by the BO+LSTM method, and even cannot meet the 30% error requirement at CM=160 cycles. In contrast, the errors of the conventional LSTM and RNN methods are larger, while the predicted RUL errors using the proposed model meet the requirements for all CM cycles. In addition, the evaluation of the overall performance of the model shows that as an improvement on the RNN, the average relative accuracy (Y_ARA) of the LSTM method improves from 34.65% of RNN to 50.53% and the average width of prediction interval (W_API) reduces from 365.3 cycles to 272 cycles. In comparison, the BO+LSTM method has a better prediction performance. Furthermore, the Y_ARA of the proposed model improves to 90.91% and the W_API decreases to 169.3 cycles, which is the best performance among several models. Quantitative analysis shows that the proposed model improves the lifetime prediction accuracy by 13% and reduces the prediction uncertainty by 34% compared to the BO+LSTM model.
The conclusions can be drawn: (1) The BO algorithm is used to optimize the hyperparameters of the LSTM, which improves the prediction accuracy of the model. (2) The SVMD is used to extract the degraded features of the IGBT, which reduces the uncertainty and improves the accuracy of the model prediction. (3) Compared with other models, the proposed model can maintain a high prediction accuracy with less CM data and its long-term prediction performance is better.

Key words： IGBT reliability lifetime prediction mode decomposition failure distribution

收稿日期: 2022-11-29

PACS:	TN322.8
	TM46

基金资助:

国家杰出青年科学基金（52125704）和中央高校基本科研业务费专项资金（2022CDJHLW009）资助项目

通讯作者: 余瑶怡女,1994年生,博士研究生,研究方向为电力电子变流器可靠性等。E-mail：yuyaoyi@cqu.edu.cn

作者简介: 任宏宇男,1997年生,硕士研究生,研究方向为电力电子变流器可靠性。E-mail：rhy@cqu.edu.cn

引用本文:

任宏宇, 余瑶怡, 杜雄, 刘俊良, 周君洁. 基于优化长短期记忆神经网络的IGBT寿命预测模型[J]. 电工技术学报, 0, (): 131-131. Ren Hongyu, Yu Yaoyi, Du Xiong, Liu Junliang, Zhou Junjie. IGBT Lifetime Prediction Model Based on Optimized Long Short-Term Memory Neural Network. Transactions of China Electrotechnical Society, 0, (): 131-131.

链接本文:

https://dgjsxb.ces-transaction.com/CN/10.19595/j.cnki.1000-6753.tces.222231 https://dgjsxb.ces-transaction.com/CN/Y0/V/I/131

[1] Yang S, Bryant A, Mawby P, et al.An industry based survey of reliability in power electronic converters[J]. IEEE Transactions on Industry Applications, 2011, 47(3): 1441-1451.
[2] 张军, 张犁, 成瑜. IGBT 模块寿命评估研究综述[J]. 电工技术学报, 2019, 34(10): 2069-2078.
Zhang Jun, Zhang Li, Cheng Yu.Review of the lifetime evaluation for the IGBT module[J]. Transactions of China Electrotechnical Society, 2021, 36(12): 2560-2575.
[3] 曾文彬, 宋梁, 张西应, 等. 基于Coffin-Manson模型功率半导体器件可靠性评估[J].电力电子技术, 2022, 56(7): 138-140.
Zeng Wenbin, Song Liang, Zhang Xiying, et al.Reliability evaluation for power semiconductor device using Coffin-Manson model[J]. Power Electronics, 2022, 56(7): 138-140.
[4] 赖伟, 陈民铀, 冉立, 等.老化实验条件下的IGBT寿命预测模型[J].电工技术学报, 2016, 31(24): 173-180.
Lai Wei, Chen Minyou, Ran Li, et al.IGBT lifetime model based on aging experiment[J].Transactions of China Electrotechnical Society, 2016, 31(24): 173-180.
[5] Ceccarelli L, Kotecha R M, Bahman A S, et al.Mission-profile-based lifetime prediction for a SiC MOSFET power module using a multi-step condition-mapping simulation strategy[J]. IEEE Transactions on Power Electronics, 2019, 34(10): 9698-9708.
[6] Shen Yanfeng, Liivik E, Blaabjerg F, et al.Reliability evaluation of an impedance-source PV microconverter[C]//2018 IEEE Applied Power Electronics Conference and Exposition (APEC), San Antonio, TX, USA, 2018: 1104-1108.
[7] Fu S Y, Tseng Y C, Chiang K N.Study on data effect of using RNN model to predict reliability life of wafer level packaging[C]//2020 15th International Microsystems, Packaging, Assembly and Circuits Technology Conference (IMPACT), Taipei, Taiwan, China, 2020: 200-203.
[8] Mei Wenjuan, Liu Zhen, Su Yuanzhang.MRPM: multistep robust prediction machine for degradation time series projection[C]//2021 IEEE International Instrumentation and Measurement Technology Conference (I2MTC), Virtual, Glasgow, UK, 2021: 1-7.
[9] Dusmez S, Duran H, Akin B.Remaining useful lifetime estimation for thermally stressed power MOSFETs based on on-State resistance variation[J]. IEEE Transactions on Industry Applications, 2016, 52(3): 2554-2563.
[10] Celaya J, Saxena A, Saha S, et al.Prognostics of power MOSFETs under thermal stress accelerated aging using data-driven and model-based methodologies[C]//Annual Conference of the Prognostics and Health Management Society, Montreal, QC, Canada, 2011: 443-452.
[11] Baharani M, Biglarbegian M, Parkhideh B, et al.Real-time deep learning at the edge for scalable reliability modeling of Si-MOSFET power electronics converters[J]. IEEE Internet of Things Journal, 2019, 6(5): 7375-7385.
[12] 李畸勇, 张伟斌, 赵新哲,等. 改进鲸鱼算法优化支持向量回归的光伏最大功率点跟踪[J]. 电工技术学报, 2021, 36(9): 1771-1781.
Li Jiyong, Zhang Weibin, Zhao Xinzhe, et al.Global maximum power point tracking for PV array based on support vector regression optimized by improved whale algorithm[J]. Transactions of China Electrotechnical Society, 2021, 36(9): 1771-1781.
[13] Zheng Shuai, Ristovski K, Farahat A, et al.Long short-term memory network for remaining useful life estimation[C]//2017 IEEE International Conference on Prognostics and Health Management (ICPHM), Dallas, TX, USA, 2017: 88-95.
[14] Ni Z, Lyu X, Yadav O P, et al.Overview of real-time lifetime prediction and extension for SiC power converters[J]. IEEE Transactions on Power Electronics, 2020, 35(8): 7765-7794.
[15] 唐圣学, 张继欣, 姚芳, 等. IGBT模块寿命预测方法研究综述[J]. 电源学报, 2023, 21(1): 177-194.
Tang Shengxue, Zhang Jixin, Yao Fang, et al.An overview of lifetime prediction methods for IGBT power module[J]. Journal of Power Supply, 2023, 21(1): 177-194.
[16] 石怀涛, 尚亚俊, 白晓天, 等. 基于贝叶斯优化的SWDAE-LSTM滚动轴承早期故障预测方法研究[J]. 振动与冲击, 2021, 40(18): 286-297.
Shi Huaitao, Shang Yajun, Bai Xiaotian, et al.Early fault prediction method combining SWDAE and LSTM for rolling bearings based on Bayesian optimization[J]. Journal of Vibration and Shock, 2021, 40(18): 286-297.
[17] 姚艳, 曹健. 一种成本有效的面向超参数优化的工作流执行优化方法[J]. 计算机集成制造系统, 2020, 26(6): 1628-1635.
Yao Yan, Cao Jian.Cost-effective workflow execution strategy for hyperparameter search[J]. Computer Integrated Manufacturing Systems, 2020, 26(6): 1628-1635.
[18] 葛建文, 黄亦翔, 陶智宇, 等. 基于Transformer模型的IGBT剩余寿命预测[J]. 半导体技术, 2021, 46(4): 316-323.
Ge Jianwen, Huang Yixiang, Tao Zhiyu, et al.Residual useful life prediction of IGBTs based on transformer model[J]. Semiconductor Technology, 2021, 46(4): 316-323.
[19] 白梁军, 黄萌, 饶臻, 等. 基于GARCH模型的IGBT寿命预测[J]. 中国电机工程学报, 2020, 40(18): 5787-5796.
Bai Liangjun, Huang Meng, Rao Zhen, et al.Lifetime prediction of IGBT based on GARCH model[J]. Proceedings of the CSEE, 2020, 40(18): 5787-5796.
[20] 高伟, 张琼洁, 李长留, 等. 基于LSTM网络的牵引变流器IGBT故障预测方法研究[J]. 电子器件, 2020, 43(4): 804-808.
Gao Wei, Zhang Qiongjie, Li Changliu, et al.A fault prediction method of IGBT in traction converter based on LSTM[J]. Chinese Journal of Electron Devices, 2020, 43(4): 804-808.
[21] 王飞, 黄涛, 杨晔, 等. 基于Stacking多模型融合的IGBT器件寿命的机器学习预测算法研究[J]. 计算机科学, 2022, 49(增刊1): 784-789.
Wang Fei, Huang Tao, Yang Ye, et al.Study on machine learning algorithms for life prediction of IGBT devices based on stacking multi-model fusion[J]. Computer Science, 2022, 49(S1): 784-789.
[22] Celaya J, Wysocki P, Goebel K. IGBT accelerated aging data set, NASA prognostics data repository[DB/OL]. NASA Ames Research Center, Moffett Field, CA, 2009, https://www.nasa.gov/content/prognostics-center-of-excellence-data-set-repository
[23] 石耀霖, 李林芳, 程术. 运用LSTM神经网络对川滇地区的地震中期预报——回溯性预测2008年汶川M_S8.0地震的探索[J]. 中国科学院大学学报, 2022, 39(1): 1-12.
Shi Yaolin, Li Linfang, Cheng Shu.Application of LSTM neural network for intermediate-term earthquake prediction: retrospective prediction of 2008 Wenchuan MS8. 0 Earthquake[J]. Journal of University of Chinese Academy of Sciences, 2022, 39(1): 1-12.
[24] 罗仁泽,李阳阳.一种基于RUnet卷积神经网络的地震资料随机噪声压制方法[J].石油物探, 2020, 59(1): 51-59.
Luo Renze,Li Yangyang.Random seismic noise attenuation based on RUnet convolutional neural network[J]. Geophysical Prospecting for Petroleum, 2020, 59(1): 51-59.
[25] 于永进, 姜雅男, 李长云. 基于鲸鱼优化-长短期记忆网络模型的机-热老化绝缘纸剩余寿命预测方法[J]. 电工技术学报, 2022, 37(12): 3162-3171.
Yu Yongjin, Jiang Yanan, Li Changyun.Prediction method of insulation paper remaining life with mechanical-thermal synergy based on whale optimization algorithm-long-short term memory model[J]. Transactions of China Electrotechnical Society, 2022, 37(12): 3162-3171.
[26] 王琛, 王颖, 郑涛, 等. 基于ResNet-LSTM网络和注意力机制的综合能源系统多元负荷预测[J]. 电工技术学报, 2022, 37(7): 1789-1799.
Wang Chen, Wang Ying, Zheng Tao, et al.Multi-energy load forecasting in integrated energy system based on ResNet-LSTM network and attention mechanism[J]. Transactions of China Electrotechnical Society, 2022, 37(7): 1789-1799.
[27] 葛磊蛟, 赵康, 孙永辉, 等. 基于孪生网络和长短时记忆网络结合的配电网短期负荷预测[J]. 电力系统自动化, 2021, 45(23): 41-50.
Ge Leijiao, Zhao Kang, Sun Yonghui, et al.Short-term load forecasting of distribution network based on combination of siamese network and long short-term memory network[J]. Automation of Electric Power Systems, 2021, 45(23): 41-50.
[28] 杨晶显, 张帅, 刘继春, 等. 基于VMD和双重注意力机制LSTM的短期光伏功率预测[J]. 电力系统自动化, 2021, 45(3): 174-182.
Yang Jingxian, Zhang Shuai, Liu Jichun, et al.Short-term photovoltaic power prediction based on variational mode decomposition and long short-term memory with dual-stage attention mechanism[J]. Automation of Electric Power Systems, 2021, 45(3): 174-182.
[29] 黄凯, 丁恒, 郭永芳, 等. 基于数据预处理和长短期记忆神经网络的锂离子电池寿命预测[J]. 电工技术学报, 2022, 37(15): 3753-3766.
Huang Kai, Ding Heng, Guo Yongfang, et al.Prediction of remaining useful life of lithium-ion battery based on adaptive data preprocessing and long short-term memory network[J]. Transactions of China Electrotechnical Society, 2022, 37(15): 3753-3766.
[30] Tamssaouet F.Towards system-level prognostics: modeling, uncertainty propagation and system remaining useful life prediction[D]. Toulouse: Institut National Polytechnique de Toulouse, 2020.
[31] 代杰杰, 宋辉, 杨祎, 等. 基于栈式降噪自编码器的输变电设备状态数据清洗方法[J]. 电力系统自动化, 2017, 41(12): 224-230.
Dai Jiejie, Song Hui, Yang Yi, et al.Cleaning method for status data of power transmission and transformation equipment based on stacked denoising autoencoders[J]. Automation of Electric Power Systems, 2017, 41(12): 224-230.
[32] 史佳琪, 马丽雅, 李晨晨, 等.基于串行-并行集成学习的高峰负荷预测方法[J]. 中国电机工程学报, 2020, 40(14): 4463-4472, 4726.
Shi Jiaqi, Ma Liya, Li Chenchen, et al.Daily peak load forecasting based on sequential-parallel ensemble learning[J]. Proceedings of the CSEE, 2020, 40(14): 4463-4472, 4726.
[33] 姚艳,曹健. 一种成本有效的面向超参数优化的工作流执行优化方法[J]. 计算机集成制造系统, 2020, 26(6): 1628-1635.
Yao Yan, Cao Jian.Cost‐effective workflow execution strategy for hyperparameter search[J]. Computer Integrated Manufacturing Systems, 2020, 26(6): 1628-1635.
[34] Nazari M, Sakhaei S M.Successive variational mode decomposition[J]. Signal Processing, 2020, 174: 107610.
[35] 张鑫. 基于信号处理的牵引逆变器系统故障诊断算法研究[D]. 成都: 西南交通大学, 2021.
[36] 李亚茹, 张宇来, 王佳晨. 面向超参数估计的贝叶斯优化方法综述[J]. 计算机科学, 2022, 49(1): 86-92.
Li Yaru, Zhang Yulai, Wang Jiachen.Survey on bayesian optimization methods for hyper-parameter tuning[J]. Computer Science, 2022, 49(1): 86-92.
[37] 黄梓欣, 林湘宁, 马啸, 等. 含风电继电保护应用中的电流互感器饱和电流重构方法[J]. 电工技术学报, 2022, 37(19): 4823-4834.
Huang Zixin, Lin Xiangning, Ma Xiao, et al.Reconstruction method of saturation current of current transformer in relay protection application related to wind power[J].Transactions of China Electrotechnical Society, 2022, 37(19): 4823-4834.
[38] 孟晓承, 韩学山, 许易经, 等. SF₆高压断路器机械故障概率的非精确条件估计[J]. 电工技术学报, 2019, 34(4): 693-702.
Meng Xiaocheng, Han Xueshan, Xu Yijing, et al.Imprecise estimation for conditional mechanical outage probabilities of SF₆ high voltage circuit breakers[J]. Transactions of China Electrotechnical Society, 2019, 34(4): 693-702.
[39] Wang Xiang, Wei Weiwei, Zhang Yanhui, et al.A data-driven lifetime prediction method for thermal stress fatigue failure of power MOSFETs[J]. Energy Reports, 2022, 8(15): 467-473.
[40] Su Xiaohong, Wang Shuai, Pecht M, et al.Prognostics of lithium-ion batteries based on different dimensional state equations in the particle filtering method[J]. Transactions of the Institute of Measurement and Control. 2017, 39(10): 1537-1546.
[41] 郭稳. 功率MOSFET剩余使用寿命预测方法及热疲劳建模研究[D]. 南昌: 华东交通大学, 2020.
[42] Chen W, Zhang L, Pattipati K, et al.Data-driven approach for fault prognosis of SiC MOSFETs[J]. IEEE Transactions on Power Electronics, 2020, 35(4): 4048-4062.