为提升柴油机故障诊断的精度与鲁棒性，本文提出一种基于卷积神经网络（Convolutional Neural Network，CNN）与D-S证据理论的融合诊断模型。该方法首先利用格拉姆角场（Gramian Angular Field，GAF）将一维振动信号转换为二维图像，再构建多种结构差异化的CNN子模型以提取故障特征并完成初步分类。在此基础上，选取性能最优的子模型作为基分类器，并引入D-S证据理论对各分类器的输出结果进行融合推理，从而提高诊断结果的可信度与稳定性。仿真实验结果表明，与传统CNN及其他智能诊断方法相比，本文所提模型在船用柴油机故障识别中表现出更高的准确率和更强的抗干扰能力，为船舶柴油机的状态监测与健康管理提供了有效技术支撑。
To enhance the accuracy and robustness of fault diagnosis, this study proposes a fusion model that integrates convolutional neural networks (CNNs) with Dempster–Shafer (D-S) evidence theory. The approach first converts one-dimensional vibration signals into two-dimensional images using the Gramian Angular Field (GAF) method. Multiple CNN sub-models with different architectures are then constructed to extract fault features and perform preliminary classification. The best-performing sub-models are selected as base classifiers, and their outputs are subsequently fused through the D-S evidence theory to improve the credibility and stability of diagnostic decisions. Simulation results demonstrate that, compared with traditional CNNs and other intelligent algorithms, the proposed model achieves higher accuracy and stronger anti-interference capability in marine diesel engine fault recognition, thereby providing effective technical support for condition monitoring and health management of ship diesel engines.
2026,48(8): 83-89 收稿日期：2025-8-25
DOI：10.3404/j.issn.1672-7649.2026.08.013
分类号：U664
作者简介：罗自来(1981-),男,博士,副研究员,研究方向为装备监测与故障诊断
参考文献：
[1] 殷文慧, 王飞, 孙强, 等. 船用柴油机人工智能故障诊断应用综述[J]. 舰船科学技术, 2023, 45(24): 122-127 YIN W H, WANG F, SUN Q, et al. A review of the application of artificial intelligence in fault diagnosis for marine diesel engines[J]. Ship Science and Technology, 2023, 45(24): 122-127
[2] 宋凯, 黄盟, 尤健, 等. 基于改进残差卷积网络的柴油机故障诊断方法[J]. 内燃机工程, 2023, 44(5): 66-73 SONG K, HUANG M, YOU J, et al. A fault diagnosis method for diesel engines based on an improved residual convolutional network[J]. Internal Combustion Engine Engineering, 2023, 44(5): 66-73
[3] 李少康, 陈龙, 陈辉, 等. 基于GAF-CNN的柴油机振动信号故障诊断[J]. 武汉理工大学学报, 2023, 47(4): 648-653 LI S K, CHEN L, CHEN H, et al. Fault diagnosis of diesel engine vibration signals based on GAF-CNN[J]. Journal of Wuhan University of Technology, 2023, 47(4): 648-653
[4] 蔡一杰, 陈俊杰, 王君, 等. 基于遗传算法优化支持向量机的柴油机气门漏气故障智能诊断方法[J]. 内燃机工程, 2022, 43(2): 71-76 CAI Y J, CHEN J J, WANG J, et al. Intelligent fault diagnosis of diesel engine valve leakage using a genetic algorithm-optimized support vector machine[J]. Internal Combustion Engine Engineering, 2022, 43(2): 71-76
[5] PAN H H, et al. An improved bearing fault diagnosis method using one-dimensional CNN and LSTM[J]. Journal of Mechanical Engineering, 2018, 7–8: 443–452.
[6] 张康, 陶建峰, 覃程锦, 等. 随机丢弃和批标准化的深度卷积神经网络柴油机失火故障诊断[J]. 西安交通大学学报, 2019, 53(8): 159-166 ZHANG K, TAO J F, QIN C J, et al. Diesel engine misfire fault diagnosis of a deep convolutional neural network with random dropout and batch normalization[J]. Journal of Xi'an Jiaotong University, 2019, 53(8): 159-166
[7] 张永祥, 王宇, 姚晓山. 基于加窗和卷积神经网络的柴油机拉缸故障诊断[J]. 车用发动机, 2019(6): 84-89
[8] 路鹏. 基于改进CNN 的船舶柴油机故障诊断方法研究[D]. 武汉: 武汉理工大学, 2021.
