针对船舶电机在复杂运行环境中易出现多种故障、且工况变化对故障特征提取造成干扰的问题，构建了一种基于连续小波变换（Continuous Wavelet Transform，CWT）与并行双通道卷积神经网络（Parallel Dual-Channel CNN，PDCNN）相结合的混合工况故障诊断模型。该方法将原始振动信号分别进行一维特征提取和二维CWT时频图变换，形成双模态输入数据，对数据提取多尺度特征后使用PDCNN进行特征融合与分类。测试结果表明，所提出模型在混合工况下的故障识别准确率达92.10%，相比仅使用一维信号或二维图像输入的模型准确率分别提高了16.88%与6.28%。同时，不同故障类型的特征区分度在t分布随机邻域嵌入（t-distributed Stochastic Neighbor Embedding，t-SNE）可视化中表现明显。研究结果说明，融合CWT与PDCNN结构能够有效提升电机在复杂工况下的故障诊断精度与鲁棒性，具有较强的工程应用潜力。
A hybrid fault diagnosis model based on the combination of continuous wavelet transform (CWT) and parallel dual channel convolutional neural network (PDCNN) is constructed to address the problem of multiple faults in ship motors in complex operating environments and the interference caused by changes in operating conditions on fault feature extraction. This method performs one-dimensional feature extraction and two-dimensional CWT time-frequency transform on the original vibration signal to form bimodal input data. After extracting multi-scale features from the data, PDCNN is used for feature fusion and classification. The test results show that the proposed model achieves a fault recognition accuracy of 92.10% under mixed operating conditions, which is 16.88% and 6.28% higher than models that only use one-dimensional signals or two-dimensional image inputs, respectively. Meanwhile, the feature discrimination of different fault types is evident in the t-distributed stochastic neighbor embedding (t-SNE) visualization. The research results indicate that the fusion of CWT and PDCNN structures can effectively improve the fault diagnosis accuracy and robustness of motors under complex working conditions, and has strong potential for engineering applications.
2026,48(2): 102-107 收稿日期：2025-5-18
DOI：10.3404/j.issn.1672-7649.2026.02.017
分类号：U664
基金项目：国家重点研发计划项目（2019YFE0104600）；国家自然科学基金资助项目（51909200）
作者简介：尚垣吉(2001-),男,硕士研究生,研究方向为系统仿真与智能诊断等
参考文献：
[1] 汤蕴璆. 电机学[M]. 北京: 机械工业出版社, 2014.
[2] 周海军, 王超伟, 周国敬, 等. 基于随机森林的离心泵滚动轴承故障诊断[J]. 中国舰船研究, 2020, 15(3): 129-135.
ZHOU H J, WANG C W, ZHOU G J, et al. Fault diagnosis of centrifugal pump rolling bearings based on random forest[J]. China Shipbuilding Research, 2020, 15(3): 129-135.
[3] CHOUDHARY A, MISHRA R K, FATIMA S, et al. Multi-input CNN based vibro-acoustic fusion for accurate fault diagnosis of induction motor[J]. Engineering Applications of Artificial Intelligence, 2023, 120: 1-12.
[4] 张冰, 姜培刚, 林天然. 基于MSST和双通道CNN技术的变转速轴承故障诊断研究[J]. 机电工程, 2021, 38(9): 1145-1151.
ZHANG B, JIANG P G, LIN T R. Research on fault diagnosis of variable speed bearings based on MSST and dual channel CNN technology[J]. Mechanical and Electrical Engineering, 2021, 38(9): 1145-1151.
[5] 唐红涛, 李冰, 高晟博. 基于双通道特征融合的CNN-LSTM轴承故障诊断方法[J]. 数字制造科学, 2022, 20(4): 253-257
TANG H T, LI B, GAO S B. CNN-LSTM bearing fault diagnosis method based on dual channel feature fusion[J]. Digital Manufacturing Science, 2022, 20(4): 253-257
[6] SUN J, GU X, HE J, et al. A robust approach of multi-sensor fusion for fault diagnosis using convolution neural network[J]. Journal of Dynamics, Monitoring and Diagnostics, 2022: 103–110.
[7] QIAN L, LI B, CHEN L. CNN-based feature fusion motor fault diagnosis[J]. Electronics, 2022, 11(17): 1-19.
[8] YAN R Q, ROBERT X G, CHEN X F . Wavelets for fault diagnosis of rotary machines: a review with applications[J]. Signal Processing, 2014, 961(Parta): 15.
[9] CHEN T, SUN L X, YU H B, et al. Online Fe grade monitoring of iron ore slurry by Morse wavelet transform and lightweight convolutional neural network based on LIBS[J]. Spectrochimica Acta Part B: Atomic Spectroscopy, 2023, 210.
[10] 尚前明, 蒋婉莹, 周毅, 等. 基于MPDCNN的强噪声环境下船舶电力推进器齿轮箱故障诊断方法[J]. 中国舰船研究, 2025, 20(2): 30-38
SHANG Q M, JIANG W Y, ZHOU Y, et al. Fault diagnosis method for ship electric propulsion gearbox in strong noise environment based on MPDCNN[J]. China Journal of Ship, 2025, 20(2): 30-38
