特征降维下基于LSSA-SVM的转子系统故障诊断模型  

Rotor system fault diagnosis model based on LSSA-SVM under feature dimensionality reduction

作　　者：史宗帅 亚森江·加入拉 崔鹏飞 靳鹏飞 SHI Zongshuai;YASENJIANG Jiarula;CUI Pengfei;JIN Pengfei(School of Mechanical Engineering,Xinjiang University,Urumqi 830017,China)

机构地区：[1]新疆大学机械工程学院,新疆乌鲁木齐830017

出　　处：《机电工程》2025年第3期463-471,500,共10页Journal of Mechanical & Electrical Engineering

基　　金：国家自然科学基金资助项目(52065065);国家教育部博士点基金资助项目(2019690007)。

摘　　要：针对有噪声环境下轴承转子系统的故障特征难以有效提取,且转子系统故障诊断的准确率较低的问题,提出了一种基于Levy飞行策略改进的麻雀搜索算法(LSSA)优化支持向量机(SVM),结合主成分分析(PCA)特征降维的转子故障诊断方法(模型)。首先,采用小波分析技术对原始的转子振动信号进行了去噪处理,通过提取信号的时域特征以精确表征不同的转子故障状态,确保了该特征在噪声干扰下仍能清晰反映故障模式;然后,采用PCA对所提取的高维特征进行了降维处理,有效减少了冗余信息和噪声干扰,保留了最具代表性的关键特征,从而提高了特征提取的效率与诊断的可靠性;最后,设计了Levy飞行策略,对SSA进行了改进,得到了改进后的麻雀搜索算法(LSSA),以优化SVM的参数选择,进一步提升了分类器的泛化能力,利用改进的算法增强了该模型在复杂、有噪声环境下的诊断性能。研究结果表明:通过在多个含噪声的转子故障数据集上进行实验,该方法的故障诊断准确率达到了98.5%,相较于传统诊断方法,其具有更强的鲁棒性和较高的诊断精度,特别是在有噪环境中的优势更为明显。该方法有效解决了噪声干扰对故障诊断精度的影响问题,显著提高了转子故障诊断的准确性和稳定性,为实际工程中的转子故障诊断提供了一种有效的解决方案。Aiming at the problem that it is difficult to effectively extract fault features of bearings and rotors in noisy environments,and the diagnosis accuracy is relatively low,a rotor fault diagnosis method based on Levy flight strategy improved sparrow search algorithm(LSSA)optimized support vector machine(SVM)combined with principal component analysis(PCA)for feature dimensionality reduction was proposed.Firstly,wavelet analysis was applied to the raw rotor vibration signals for denoising,allowing time-domain features to be extracted that accurately represent various rotor fault states,even under significant noise interference,ensuring that the features retain critical fault information.Then,PCA was used to reduce the dimensionality of the extracted high-dimensional features,effectively filtering out redundant information and noise,while preserving the most representative and significant features,thus improving the efficiency of feature extraction and the reliability of the diagnosis process.Finally,a Levy flight strategy was introduced to enhance the SSA to build the improved sparrow search algorithm(LSSA)for optimizing SVM parameters,further improving the generalization capability of the classifier and boosting the model s diagnostic performance in noisy and complex environments.The experimental results show that:after testing on multiple rotor fault datasets with noise,the proposed method achieves a diagnostic accuracy of 98.5%,demonstrating superior robustness and higher diagnostic precision compared to traditional methods,with notable advantages in noisy environments.In conclusion,the proposed method effectively addresses the impact of noise interference on diagnostic accuracy,significantly improving the accuracy and stability of rotor fault diagnosis,and offering an effective solution for rotor fault detection in real-world engineering applications.

关 键 词：轴承故障诊断 莱维飞行 改进的麻雀搜索算法 支持向量机 主成分分析 主成分分析特征降维 小波阈值函数去噪 

分 类 号：TH133.3[机械工程—机械制造及自动化]