|
应用LDP降维的电机转子系统故障诊断研究
李媛媛，谢文彬 2026/3/10 17:58:47
江苏联合职业技术学院淮安生物工程分院，机电工程系，江苏淮安 223200
摘要：诊断电机转子系统故障中主成份分析(PCA)存在较低准确率。为了进一步提高振动信号特征提取能力，建立了一种基于局部判别投影(LDP)降维方法的转子系统故障状态诊断方法。搭建了转子振动实验台，并通传感器过对转动不平衡性的振动信号进行提取以及降维分析。降维效果方面，LDP算法较其它方法具有明显优势，实现了更优的可分性，显著提升数据处理效率。相对其他方法，LDP能够充分利用判别特征和局部拓扑关系，显著增强故障特征集合区分能力，能够获得更优越的故障识别性能。
关键词：电机转子系统；局部判别投影；信号降维；故障诊断
中图分类号：TH165
Research on Fault Diagnosis of Motor Rotor System Using LDP Dimension Reduction
Li Yuanyuan, Xie Wenbin
Department of Mechanical and Electrical Engineering, Huai ’an Branch of Jiangsu Vocational College of Technology, Huai ’an 223200, Jiangsu
Abstract: Principal Component Analysis (PCA) has a relatively low accuracy rate in diagnosing faults in motor rotor systems. In order to further improve the feature extraction ability of vibration signals, a fault state diagnosis method for rotor systems based on the local Discriminant projection (LDP) dimension reduction method is established. A rotor vibration test bench was set up, and the vibration signals of rotational imbalance were extracted and dimensionality reduction analysis was conducted through sensors. In terms of the dimension reduction effect, the LDP algorithm has obvious advantages over other methods, achieving better separability and significantly improving the efficiency of data processing. Compared with other methods, LDP can make full use of discriminative features and local topological relationships, significantly enhance the discrimination ability of fault feature sets, and achieve superior fault identification performance.
Key words: Motor rotor system; Local discriminant projection; Signal dimension reduction; Fault diagnosis
1 引言
针对电机在运行期间形成的故障进行检测时，运行参数采集的完整性直接影响诊断结果可靠性[1]。通过分析设备运转阶段产生的振动信号能够高效识别机械系统本质运动特征的变化，这种监测方法不仅可以获得优异识别精度，操作流程较简便，目前已成为各类电机转子系统故障诊断的主流方法[32]。
当前关于数据降维技术的有效性研究正受到越来越多学者的关注，尤其是基于大数据分析的相关方法获得了明显重视[3]。蔡兆龙等[4]提出种融合纵向空间特征提取和注意力机制的故障诊断模型，实现多个时刻下振动信号时序特征与故障类别之间的映射，提升转子振动故障诊断的准确率。樊红卫等[5]提出角域重（未完，下一页）
附件下载：应用LDP降维的电机转子系统故障诊断研究
|