基于多项式滤波算法的自适应稳态检测  

Method of Adaptive Steady-state Detection Based on Polynomial Filtering

作　　者：付克昌[1] 戴连奎[1] 吴铁军[1] 

机构地区：[1]浙江大学工业控制技术国家重点实验室智能系统与决策研究所,杭州310027

出　　处：《化工自动化及仪表》2006年第5期18-22,共5页Control and Instruments in Chemical Industry

基　　金：国家高技术研究发展计划项目(2003AA412110)

摘　　要：提出一种基于多项式滤波的自适应稳态检测方法。该方法采用滑动窗口方法对当前窗口内的数据进行多项式滤波,根据滤波后的多项式的一次项系数的大小判断测量信号的稳态与非稳态,而滤波窗口长度可根据训练样本自适应地确定。与现有方法相比,基于多项式滤波的自适应稳态检测方法具有实现简单、运算速度快、适用于信号的在线自动稳态检测。通过非绝热连续反应釜的仿真验证了本方法的有效性。An adaptive polynomial filtering（APF） method for steady-state detection is presented, which adopts slide window method to make polynomial filtering for the present window＇ s data. Steady state of process variables is determined by the first-order coefficients of polynomial filtering, and the size of the filter window is decided adaptively by the training sample. Implementation of APF is easier and faster than the existed methods, and it is suited to online adaptive steady-state detection. Simulation results for a non-isothermal CSTR process demonstrate the effectiveness and efficiency of the proposed method.

关 键 词：稳态检测 多项式滤波 过程监控 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置]