基于支持向量机的直升机旋翼系统故障诊断

Fault diagnosis based on hierarchical clustering support vector machine for helicopter rotor

采用小波包消噪方法对直升机机体振动信号进行消噪,以抑制各频带内的宽带随机噪声,并对消噪后的信号进行单支重构以提取各频带能量,将频带能量作为识别旋翼故障类型的特征向量.针对支持向量机多类方法中存在拒绝分类区的缺点,将分级聚类的思想和决策树的思想结合构造了一种新的分层聚类多类支持向量机,实现了直升机旋翼系统的故障分类和诊断.实验结果表明:小波包能够有效抑制直升机振动信号中的噪声,所提出的分层聚类支持向量机简化了分类器结构,减少了分类器数量,避免了拒绝分类区的出现,并加快了训练和识别速度,正确实现了直升机旋翼系统的故障识别.

Due to weak fault signal and difficulty in extracting fault feature for helicopter rotor,the wavelet package analysis is adopted to eliminate the noise in the actual signals and to extract energy feature vector in various frequency bands.And considering the region of rejection existing in current multi-class support vector machine(SVM) classification algorithm,a new support vector machine based on hierarchical clustering and decision tree is proposed to solve the multi-class recognition problems in helicopter rotor fault diagnosis.The experiment results indicated that wavelet package can eliminate the noise in helicopter vibration signal and that the presented multi-class SVM simplified categorizer structure,avoided the region of rejection,and accelerated the training and identifying speed.