采用高阶统计和模糊聚类的阀门黏滞故障检测  

Detection of Valve Stiction Fault Using Higher-order Statistics and Fuzzy Clustering

作　　者：郑丽丽 王志国 刘飞 

机构地区：[1]江南大学,轻工过程先进控制教育部重点实验室,江苏无锡214122

出　　处：《仪表技术与传感器》2017年第10期23-28,共6页Instrument Technique and Sensor

基　　金：国家自然科学基金项目(61134007)

摘　　要：针对气动调节阀存在的黏滞非线性故障,提出了一种采用高阶统计(HOS)和模糊聚类的阀门黏滞故障检测方法。首先,利用基于高阶统计量的非高斯性指标(NGI)和非线性指标(NLI)检测控制回路是否包含非线性环节;再根据黏滞阀门的输出输出特性设计检测算法,采集回路数据进行模糊聚类,以获取的聚类中心位置特征为基础,给出黏滞存在的判断依据并量化黏滞大小。通过计算机仿真和工业数据分析,验证了所提方法的有效性和实用性。Since stiction is a nonlinear fault generated by sticky pneumatic valve,a method using higher-order statistics（ HOS） and fuzzy clustering to detect valve stiction was proposed.Firstly,detection of nonlinearity was carried out by utilizing the non-Gaussianity index（ NGI） and the nonlinearity index（ NLI） based on HOS. Then the detection algorithm was designed according to the input-output characteristic of a sticky valve.After operating data were clustered using fuzzy clustering,stiction criterion and quantification were proposed on basis of the distribution of cluster centers.The effectiveness and practicability of the proposed method were successfully validated by using computer simulation and industrial data.

关 键 词：阀门黏滞 非线性故障 高阶统计量 模糊聚类 

分 类 号：TP214[自动化与计算机技术—检测技术与自动化装置]