基于过程模型的化工动态数据校正方法研究

李贵晓， 黄兆杰， 金思毅*

(青岛科技大学 化工学院，山东 青岛 266042)

摘要： 在工业过程中，获得准确可靠的测量数据是实现过程控制、模拟、优化和生产管理的前提条件。当测量数据中存在过失误差时，基于过程模型的卡尔曼滤波得到的校正结果准确性会降低。为了降低过失误差的影响，将鲁棒估计函数与卡尔曼滤波相结合，利用鲁棒函数的影响函数修正测量值方差，提出了基于鲁棒估计函数改进的卡尔曼滤波，并推导给出了修正方差的计算公式。动态非线性实例的应用结果表明，与传统的卡尔曼滤波相比，改进的卡尔曼滤波的过失误差校正性能有了显著提高，可有效地用于动态过程的数据校正。

关键词： 过失误差侦破； 卡尔曼滤波； 鲁棒估计函数； 动态数据校正

中图分类号： TQ 056                      文献标志码：   A

Studies on Chemical Dynamic Data Reconciliation Method Based on Kalman Filter

LI Guixiao, HUANG Zhaojie, JIN Siyi

 (College of Chemical Engineering, Qingdao University of Science and Technology, Qingdao 266042，China)

Abstract： For process industries, it is crucial for the control, simulation and management of the process to obtain high quality and reliability data. When the gross error exists in the measured data, the data correction performance of Kalman filter based on process model will decline. In order to reduce the influence of gross errors, in this paper, an improved Kalman filter based on the robust estimation function is proposed; also the calculation formula for modified Kalman filter is deduced. By the proposed method, the influence function of robust estimator is used to modify the measurement′s variance, as a result the difference between the corrected value and the measurement containing gross error increase, and the influence of gross errors is reduced. The simulation results of a nonlinear dynamic instance show that, compared with the traditional Kalman filter, the improved Kalman filter′s gross error correction performance has significantly improved, therefore, the proposed modified method can be effectively used for dynamic process data reconciliation.

Key words: gross error detection；  Kalman filter；  robust estimation function；  dynamic data reconciliation

收稿日期： 20140314

基金项目： 重质油国家重点实验室开放课题基金资助项目（201103004）.

作者简介： 李贵晓（1990—），男，硕士研究生.*通信联系人.

 文章编号：16726987（2015）03029705； DOI： 10.16351/j.16726987.2015.03.013