基于PSO-NKNN的齿轮箱故障诊断方法  

Gearbox fault diagnosis method based on particle swarm optimization-neutrosophic K-nearest neighbor

作　　者：田锟 丁云飞 陈启凡 孙钱承 TIAN Kun;DING Yunfei;CHEN Qifan;SUN Qiancheng(School of Electrical Engineering,Shanghai DianJi University,Shanghai 201306,China)

机构地区：[1]上海电机学院电气学院,上海201306

出　　处：《上海电机学院学报》2024年第6期324-330,共7页Journal of Shanghai Dianji University

基　　金：航空科学基金资助项目(20200001012015)。

摘　　要：对齿轮箱振动信号及时准确地进行故障诊断,是降低风机运维成本的关键。基于此,提出了一种基于粒子群优化的中智最近邻算法(PSO-NKNN)。首先,在初始阶段,通过采用小波包分解与重构技术对原始信号进行特征提取,以捕捉信号的能量特征;然后,引入粒子群优化算法对中智最近邻算法进行优化;最后,构建了PSO-NKNN故障诊断模型,并通过QPZZ-Ⅱ平台采集的真实数据进行实验验证。验证结果表明:该方法弥补了中智最近邻算法(NKNN)对“假”隶属度权重分配不确定的缺陷,有效地提高了分类准确度,同时提升了模型的抗噪性。Timely and accurate diagnosis of gearbox vibration signals is crucial for reducing the operational costs of wind turbines.An algorithm based on particle swarm optimization-neutrosophic K-nearest neighbor(PSO-NKNN)for this purpose is proposed in this paper.First,in the initial stage,wavelet packet decomposition and reconstruction techniques are used to extract features from the raw signal in order to capture its energy characteristics.Then,the particle swarm optimization(PSO)algorithm is introduced to optimize the nearest neighbor algorithm(NKNN).Finally,a PSO-NKNN fault diagnosis model is constructed and experimentally validated using real data collected from the QPZZ-Ⅱplatform.The experimental results show that this method compensates for the uncertainty in the weight distribution of the“false”membership degree in the NKNN,effectively improving classification accuracy while enhancing the model's noise resistance.

关 键 词：中智理论 最近邻算法 粒子群算法 故障诊断 齿轮箱 

分 类 号：TM614[电气工程—电力系统及自动化]