为实现对船舶逆变器的有效维护，确保船舶逆变器模块的正常运行，提出一种基于多尺度特征融合和降噪卷积自编码器的船舶逆变器开路故障诊断方法。可以直接对一维原始电流数据自适应地提取数据特征，降低信号内的噪声，实现端到端的故障诊断。首先，利用数据增强方法来增强数据集；其次，根据数据特点设计可以提取局部细节和整体结构信息的多尺度卷积特征融合模块，并在编码器中引入该模块，形成特征提取模型；最后，利用全连接神经网络对模型输出的数据特征进行分类，根据分类结果实现故障诊断。实验结果表明，所提出的方法具有优越的数据特征提取性能及噪声鲁棒性能，可以实现船舶逆变器开关器件开路故障诊断。
In order to realize effective maintenance of marine inverters and ensure the normal operation of marine inverter modules, the paper proposes an open-circuit fault diagnosis method for marine inverter based on multi-scale feature fusion and noise-reducing convolutional auto-encoder. Data features can be extracted adaptively and directly from the one-dimensional raw current data to reduce the noise within the signal and realize end-to-end fault diagnosis. Firstly, data enhancement methods are utilized to augment the dataset. Secondly, a multi-scale convolutional feature fusion module is designed to extract both local details and overall structural information according to the data characteristics, and the module is introduced into the encoder to form a feature extraction model. Finally, the fully connected neural network is used to classify the data features from the model, and fault diagnosis is implemented based on the classification results. The experimental results show that the proposed method has superior performance of data feature extraction and strong robustness to noise, and it can diagnose open-circuit fault of power switch in the marine inverter.
2025,47(3): 135-140 收稿日期：2024-4-26
DOI：10.3404/j.issn.1672-7649.2025.03.022
分类号：U665.13
基金项目：国家自然科学基金资助项目（51779102）；福建省自然科学基金资助项目（2022J01811）
作者简介：崔博文(1966-),男,博士,教授,研究方向为电力系统状态监测与参数估计、逆变器故障诊断
参考文献：
[1] 蒋佳炜, 胡以怀, 方云虎, 等. 船舶动力装置智能故障诊断技术的应用与展望[J]. 中国舰船研究, 2020, 15(1): 56?67.
JIANG J W, HU Y H, FANG Y H, et al. Application and prospects of intelligent fault diagnosis technology for marine power system[J]. Chinese Journal of Ship Research, 2020, 15(1): 56?67.
[2] 马铭遥, 凌峰, 孙雅蓉, 等. 三相电压型逆变器智能化故障诊断方法综述[J]. 中国电机工程学报, 2020, 40(23): 7683?7699.
MA M Y, LING F, SUN Y R, et al. Review of intelligent fault diagnosis methods for three-phase voltage-mode inverters[J]. Proceedings of the CSEE, 2020, 40(23): 7683?7699.
[3] 宋威, 施伟锋, 卓金宝, 等. 多电平逆变器开关管故障诊断方法综述[J]. 微电机, 2019, 52(10): 110?117.
SONG W, SHI W F, ZHUO J B, et al. Review of fault diagnosis methods for switching tubes of multilevel inverters[J]. Micromotors, 2019, 52(10): 110?117.
[4] 申皓澜, 唐 欣, 罗毅飞, 等. 基于 CNN 的三相逆变器开路故障诊断及其样本条件分析[J]. 国防科技大学学报, 2022, 44(6): 163?172.
SHEN H L, TANG X, LUO Y F, et al. Open-circuit fault diagnosis for three-phase inverters based on CNN and analysis of its sample conditions[J]. Journal of National University of Defense Technology, 2022, 44(6): 163?172.
[5] GONG W, CHEN H, ZHANG Z, et al. A data-driven-based fault diagnosis approach for electrical power DC-DC inverter by using modified convolutional neural network with global average pooling and 2-D feature image[J]. IEEE Access, 2020, 8: 73677?73697.
[6] YANG H, MENG C, WANG C, et al. Data-driven feature extraction for analog circuit fault diagnosis using 1-D convolutional neural network[J]. IEEE Access, 2020, 8: 18305?18315.
[7] CHEN C C , LIU Z , YANG G , et al. An improved fault diagnosis using 1D-convolutional neural network model[J]. Electronics, 2021, 10(1): 59.
[8] 宋威, 林建维, 周方泽, 等. 基于改进降噪自编码器的风机轴承故障诊断方法[J]. 电力系统保护与控制, 2022, 50(10): 61?68.
SONG W, LIN J W, ZHOU F Z, et al. Wind turbine bearing fault diagnosis method based on an improved denoising AutoEncoder[J]. Power System Protection and Control, 2022, 50(10): 61?68.
[9] CHEN X, LI M, XIAO Q Y. Stacked denoise autoencoder based feature extraction and classification for hyperspectral images[J]. Journal of Sensors, 2016: 1?10.
[10] SZEGEDY C, L W, JIA Y, et al. Going deeper with convolutions[C]//2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2015.
