摘要
针对传统基于支持向量分类机(SVC)的发动机多模式故障诊断方法需要多个二类分类器的问题,提出了一种基于支持向量回归机(SVR)的多模式故障诊断方法。该方法首先应用归一化的故障数据样本和一个支持向量回归机构建一个发动机故障诊断回归模型,再对支持向量回归机的输出结果进行基于距离的聚类操作得到发动机的故障模式,诊断模型的参数向量采用一种基于Tent混沌映射的量子粒子群优化算法及样本测试集的均方根误差与平均相对误差同时最小的准则进行整定。实验结果表明,所提出的方法能够克服常规支持向量分类机多模式故障诊断方法需要多个二类分类器的缺陷,降低了建模的时间复杂度,有效地提高了发动机的故障诊断性能。
Aiming at the problem that traditional engine multi-mode fault diagnosis method utilizing support vector classification machine (SVC) requires multiple binary classifiers, a new multi-mode fault diagnosis method utilizing support vector regression machine (SVR) is proposed in this paper to improve the performance of fault diagnosis. The normalized fault data samples and a SVR are applied to construct an engine fault diagnosis regression model; then, the clustering operation based on distance is performed on the output results of the SVR to obtain the fault mode of the engine. The parameter vectors of the diagnosis model are tuned with the quantum-behaved particle swarm optimization algorithm based on Tent chaotic mapping (TCPQPSO) and the criterion that the root-mean-square error (RMSE) and mean-relative-error (MRE) of the sample test set remain minimum simultaneously. The experiment results show that the proposed method can overcome the defect that the multi-mode fault diagnosis method utilizing conventional support vector classification machine (SVC) requires multiple binary classifiers, reduce the time complexity of modeling, and effectively improve the performance of engine fault diagnosis.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2015年第9期2112-2119,共8页
Chinese Journal of Scientific Instrument
基金
江苏省基础研究计划(自然科学基金)(BK20131124)
徐州工程学院江苏省大型工程装备检测与控制重点建设实验室开放基金(JSKLEDC201212)项目资助
关键词
发动机故障诊断
支持向量回归机(SVR)
聚类
Tent混沌映射
量子粒子群优化算法
engine fault diagnosis
support vector regression machine (SVR)
clustering
Tent chaotic mapping
quantum-behaved particle swarm optimization algorithm