摘要
在工业过程控制中,气动阀门的粘滞非线性特性会导致控制回路性能下降甚至振荡。针对已有基于模糊聚类的阀门粘滞检测方法在故障检测过程中容易出现错误诊断问题,提出了改进方法,能有效识别外部干扰和阀门粘滞。首先利用模糊聚类算法对回路日常运行数据进行聚类分析得到聚类中心,根据粘滞阀门输入输出数据的分布特性,评价聚类中心的线性拟合度。然后对聚类中心所构成四边形进行凹凸性识别,再根据聚类中心的分布特征定义了一种新粘滞指标。通过给出的仿真实验,以及化工厂的两个流量控制回路故障检测实验,验证了所提方法的有效性和准确性。
In the process industry, stiction nonlinearity existing in pneumatic control valve will lead to performance deterioration even oscillation in the control loops. Since the existing methods using fuzzy clustering algorithm may easily make wrong diagnosis in the detection, an improved method was proposed in this paper, which can effectively recognize the external disturbance and valve stiction. Firstly, the routine operating data were clustered using fuzzy clustering algorithm to obtain cluster centers and the goodness-of-fit was evaluated according to the distribution characteristic of taken. On basis of the distribution of cluster centers, a new stiction index was defined. The simulations and two flow control loops of chemical industry were used to verify the validity and veracity of the improved method.
出处
《机械科学与技术》
CSCD
北大核心
2018年第2期300-305,共6页
Mechanical Science and Technology for Aerospace Engineering
基金
国家自然科学基金项目(61134007)资助
关键词
气动阀门
粘滞非线性
模糊聚类
故障检测
凹凸性
pneumatic control valve
stiction nonlinearity
fuzzy clustering algorithm
fault detection
convex-concave property
