基于输出PDF的浮选故障检测和诊断方法  

Flotation Fault Detection and Diagnosis Method Based on Output PDF

作　　者：桂卫华[1] 杜建江[1] 许灿辉[1] 阳春华[1] 

机构地区：[1]中南大学控制科学与工程,湖南长沙414000

出　　处：《控制工程》2012年第2期307-310,共4页Control Engineering of China

基　　金：国家自然科学基金资助项目(61071176),国家自然科学基金资助项目(60874069)

摘　　要：针对浮选中泡沫尺寸分布的特殊性,如非高斯分布,左偏斜,高峰值等,常规分析方法无法准确描述尺寸分布的特点,因此无法准确检测和诊断浮选过程中出现的故障。提出对泡沫尺寸分布的输出概率密度函数(PDF)的统计分析,形成了一种新的浮选过程故障检测和诊断方法。通过采用自设计的核方法逼近将输出PDF转化为动态权系数,建立带有时滞的非线性不确定性权动态模型,基于线性矩阵不等式设计得到可行的故障检测和诊断算法。通过仿真验证分析,证明此算法的有效性。结合现场浮选过程,讨论了此方法的应用前景和优势。Considering the fact that bubble size distribution in mineral flotation is found to be non-Gaussian and highly left-skewed with spike, the conventional analysis methods are unable to describe these characteristics which are significant for detecting and diagnosing process fault in flotation. It puts forward a statistical analysis on the output probability density function （PDF） of bubble size distribu- tion. Then a new method of fault detection and diagnosis is formed. By means of estimating the output PDF of bubble size using kernel method, the PDF is transformed into dynamic weight coefficients, based on which the uncertain nonlinear dynamic models with time delay are established. According to Linear matrix inequality （LMI） a feasible fault detection and diagnosis algorithm can be obtained. The simulation results verify the effectiveness of the proposed algorithm. With the flotation process, the prospects and advantages of this method is discussed.

关 键 词：输出PDF 故障检测和诊断 核方法 权函数 

分 类 号：TP27[自动化与计算机技术—检测技术与自动化装置]