摘要
讨论了基于输入空间样本向量重构的核主元分析故障诊断方法,提出了用输入空间与特征空间的距离约束关系求取输入向量重构值的原像。将这一方法应用到水轮机调节系统的故障诊断,分析了系统的故障形式,选择了检测输入向量、确定了核函数的形式和主元个数,建立了基于核主元分析的水轮机调节系统故障诊断模型。利用水轮机调节系统的非线性状态方程产生样本数据,引入导叶开度传感器故障和流量传感器故障对故障诊断过程进行仿真,仿真结果证明了该方法应用到水轮机调节系统故障诊断中的有效性。
This paper puts forth a fault diagnosis method of kernel principal component analysis ( KPCA ) based on reconstruction of input space sample vectors. The pre-image of vector reconstruction is obtained by using a constraint on the distance between the input space and feature space. This method was applied to sensors diagnosis of hydro turbine regulation and a model of KPCA fault diagnosis was developed through analyzing fault mechanism, selecting input vectors, and determining kernel function and the number of principal components. Simulations of diagnosis on guide vane sensor fault and flow sensor fault were made using sample data produced by the nonlinear state equations for turbine regulation system. Results show that the proposed method is practically feasible and effective.
出处
《水力发电学报》
EI
CSCD
北大核心
2013年第5期261-268,共8页
Journal of Hydroelectric Engineering
关键词
水力机械
故障诊断
核主元分析
原像
水轮机调节系统
hydraulic machinery
fault diagnosis
kernel principal component analysis
pre-image
regulation of hydro turbine