基于SKICA的非线性过程缓变故障检测方法研究

Ramp fault detection method based on SKICA for nonlinear process

针对非线性工业过程缓变型故障的检测问题,提出一种基于累积和核独立元分析(SKICA)的故障检测方法。通过核函数技术将观测数据从非线性空间映射到线性空间,然后对线性空间的数据应用独立元分析算法,提取观测数据中的非线性独立元。为了更好的检测过程中微小变化和缓变故障,进一步应用累积和控制图(CUSUM)的思想。建立累积和非线性独立元并以此构造统计量监控过程变化。在连续搅拌反应器(CSTR)上的仿真结果表明,SKICA方法能够比ICA方法更快的检测出非线性过程中的缓变型故障。

In order to detect ramp fault in nonlinear industrial process, a new fault detection method is proposed based on summed kernel independent component analysis （SKICA）. Measured data in original nonlinear space is mapped into linear space by kernel function technique. Independent component analysis is performed on the transformed linear data and nonlinear independent components are extracted. For fast detection of small shift and ramp fault, CUSUM charts are applied and summed nonlinear independent components are built to develop monitoring statistics. Simulation results on continuous stirred tank reactor （CSTR） system show that SKICA can detect ramp fault earlier than ICA in nonlinear process.