CAJ | 학술논문

针对实际工业过程数据中的非线性问题,研究了一种基于非线性独立元分析的多变量过程监控方法。该方法根据贝叶斯原理,构造多层感知器网络恢复过程数据,并以此建立过程的数学统计模型,对其进行实时监控。在大型工业设备仿真器TE上的应用表明了该方法的有效性,同时,在故障诊断方面也体现出了一定的优越性。
침대실제공업과정수거중적비선성문제,연구료일충기우비선성독립원분석적다변량과정감공방법。해방법근거패협사원리,구조다층감지기망락회복과정수거,병이차건립과정적수학통계모형,대기진행실시감공。재대형공업설비방진기TE상적응용표명료해방법적유효성,동시,재고장진단방면야체현출료일정적우월성。
Considering the nonlinear characteristic of date in real industry processes,a multivariable process monitoring method based on nonlinear independent component analysis(NICA) is presented.With the help of Bayesian theorem,process data can be reconstructed by establishing multi-layer perceptrons,and statistical model of process in mathematics can be given for monitoring in real time.The proposed method is applied to the Tennessee-Eastman(TE) process.Simulation results show its availability and advantage in th...