基于形态梯度变换和改进支持向量机的多普勒雷达天线机械故障诊断方法  

Doppler weather radar machinery fault diagnosis based on the mathematical morphological gradient transformation and the revised SVM

作　　者：甄树勇[1] 蒋涛[1] 梁如意[1] 王硕飞[1] 靳松[2] 

机构地区：[1]河北省气象技术装备中心,河北石家庄050021 [2]邯郸市气象局,河北邯郸056001

出　　处：《河北省科学院学报》2013年第3期5-11,50,共8页Journal of The Hebei Academy of Sciences

摘　　要：提出了基于形态梯度变换和改进支持向量机的多普勒雷达天线机械故障诊断方法。首先基于数学形态梯度对雷达天线机械故障信号进行滤波,消除噪声因素的影响。然后,针对现有支持向量机投票策略在构造多类分类器的过程中存在的部分区域不可分的问题,提出基于Beta映射关系函数的贝叶斯优化投票策略支持向量机模型,建立基于Beta映射关系函数的贝叶斯先验分布和后验分布之间的联系,对后验分布进行精确的估计,实现对不可分区域数据的有效分类。通过对采集到的某型号多普勒雷达天线的轴承故障信号的有效分类,表明了本文所提方法的准确性。A novel method based on the mathematical morphological gradient transformation and the revised SVM for Doppler weather radar machinery fault diagnosis is proposed.Firstly use the mathematical morphological gradient to remove the noise.Then,aiming at the existing problem of the voting strategy in the construction of multi-class SVM,which could cause an inseparable part of the region,this article proposed an optimal voting strategy by Bayesian.create mapping functions based on Beta prior distribution,and link it with the posterior distribution,so as to achieve the sub-regional been separated.Through the classification of bearing fault signals of a type of Doppler radar,and compared with the result of the classical multiclass SVM,indicating the accuracy of the proposed method.

关 键 词：形态学梯度 贝叶斯 投票策略 支持向量机 故障诊断 

分 类 号：TH165.3[机械工程—机械制造及自动化] TN911.2[电子电信—通信与信息系统]