欢迎访问《兵工学报》官方网站,今天是 2025年6月15日 星期日

兵工学报

• •    下一篇

基于改进CNN-LSTM的开关闩故障诊断方法

高振华1,秦奋起2,王琳琳2,于存贵1*()   

  1. (1. 南京理工大学 机械工程学院, 江苏 南京210094; 2. 中国船舶重工集团公司第七一三研究所, 河南 郑州 450015)
  • 收稿日期:2024-09-09 修回日期:2025-03-25
  • 通讯作者: yu-cungui@njust.edu.cn

Fault diagnosis method of Breechblock Opening-closing Mechanism based on improved CNN-LSTM

GAO Zhenhua1, QIN Fenqi2, WANG Linlin2, YU Cungui1*()   

  1. (1.School of Mechanical Engineering, Nanjing University of Science and Technology, Nanjing 210094, Jiangsu, China; 2. 713th Research Institute of China Shipbuilding Industry Corporation, Zhengzhou 450015, Henan, China )
  • Received:2024-09-09 Revised:2025-03-25

摘要: 针对某舰炮开关闩机构工况复杂多变、工作环境恶劣,导致故障类型检测困难等问题,提出了一种基于麻雀优化算法(SSA)的格拉姆角场结合卷积神经网络和长短期记忆神经网络(GAF-CNN-LSTM)的开关闩故障诊断方法。首先,采集开关闩故障原始信号并进行预处理,通过时频分析法和格拉姆角场法建立一维时序数据和二维图像故障数据集;然后,将故障数据集分别输入到LSTM和CNN通道中,利用CNN强大的空间特征提取能力和LSTM挖掘数据时序特征能力进行特征提取,并将二者得到的特征信息进行融合,在全连接层和激活函数的作用下输出诊断结果;最后,通过SSA优化算法对GAF-CNN-LSTM网络结构中的超参数进行优化,提高模型的诊断精度和适用性。经测试数据验证:提出的SSA-GAF-CNN-LSTM故障诊断模型不仅可以更精准地诊断开关闩机构故障类型,而且具有更强的泛化能力及抗干扰能力,有效地提高了开关闩机构故障诊断的性能。

关键词: 故障诊断, 开关闩机构, 深度学习算法, 群体智能优化算法, 格拉姆角场

Abstract: In order to solve the problems of complex and changeable working conditions and harsh working environment of a Breechblock Opening-closing Mechanism of naval gun, which lead to the difficulty of fault type detection, a Breechblock Opening-closing Mechanism fault diagnosis method of Gramian angular field combined with convolutional neural network and long short-term memory neural network (GAF-CNN-LSTM) based on sparrow optimization algorithm (SSA) was proposed. Firstly, the original signal of the Breechblock Opening-closing Mechanism fault was collected and preprocessed, and the one-dimensional time series data and two-dimensional image fault dataset were established by the time-frequency analysis method and the Gramian angular field method; Then, the fault dataset was input into the LSTM and CNN channels respectively, and the powerful spatial feature extraction ability of CNN and the time series feature ability of LSTM mining data were used to extract features, and the feature information obtained by the two was fused to output the diagnostic results under the action of the fully connected layer and activation function. Finally, the SSA optimization algorithm is used to optimize the hyperparameters in the GAF-CNN-LSTM network structure to improve the diagnostic accuracy and applicability of the model. Verified by the test data: the proposed SSA-GAF-CNN-LSTM fault diagnosis model can not only diagnose the fault type of the Breechblock Opening-closing Mechanism of naval gun more accurately, but also has stronger generalization ability and anti-interference ability, which effectively improves the fault diagnosis performance of the Breechblock Opening-closing Mechanism.

Key words: fault diagnosis, Breechblock Opening-closing Mechanism, deep learning algorithms, swarm intelligence optimization algorithm, Gramian angular field