摘要
面向重复使用火箭发动机的状态监测与故障诊断需求,针对振动信号的非平稳性和难以提取有效故障特征的问题,提出一种基于变分模态分解(Variational Mode Decomposition, VMD)和模糊C均值聚类(Fuzzy C-Means, FCM)的状态监测方法。采用优化VMD算法自适应地将振动信号分解为多个本征模态分量(Intrinsic Mode Function, IMF),根据加权相关样本熵最大准则选取关键IMF分量;利用t分布随机近邻嵌入(t-SNE)对关键IMF分量的多维时域、频域特征降维,得到特征向量矩阵;利用模糊C均值聚类算法实现发动机工作状态的监测。将该方法应用于发动机涡轮泵工作状态监测,结果表明其能够提取振动信号关键特征,准确识别涡轮泵工作状态,测试集识别准确率达92.50%,为火箭发动机状态监测与故障诊断提供了理论支撑。
Targeting the status monitoring and fault diagnosis requirements for reusable rocket engines,addressing the issues of non-stationary vibration signals and difficulty in extracting effective fault features,a method for state monitoring based on variational mode decomposition(VMD)and fuzzy C-means(FCM)clustering is proposed.The optimized VMD algorithm was adopted to adaptively decompose the vibration signal into multiple intrinsic mode functions(IMF),and key IMF components were selected based on the weighted correlation sample entropy maximum criterion;the time-domain and frequency-domain feature dimensionality reduction of key IMF components using t-distributed stochastic neighbor embedding(t-SNE)was employed to obtain the feature vector matrix,and the fuzzy center means clustering algorithm was used to monitor the working status of the engine.The method was applied to the monitoring of the working status of the tur-bopump,and the results showed that it can extract key features of vibration signals and accurately identify the working status of the turbopump.The recognition accuracy of the test set reached 92.50%,providing theoretical support for status monitoring and fault diagnosis of rocket engines.
作者
敖一峰
李洪
张金刚
黄辉
AO Yifeng;LI Hong;ZHANG Jingang;HUANG Hui(Beijing Institute of Astronautical Systems Engineering,Beijing 100076,China;China Aerospace Science and Technology Corporation,Beijing 100048,China)
出处
《测试技术学报》
2024年第5期527-534,551,共9页
Journal of Test and Measurement Technology
基金
科技部国家重点研发计划(2021YFB3203300)。
关键词
火箭发动机
涡轮泵
状态监测
振动信号
变模态分解
模糊均值聚类
rocket engine
turbopump
condition monitoring
vibration signal
variational mode decompo-sition
fuzzy C-means