基于改进CVA算法的系统辨识方法  

System Identification Method Based on Improved CVA Algorithm

作　　者：臧春华 高伟 ZANG Chun-hua;GAO Wei(School of Information Engineering,Shenyang University of Chemical Technology)

机构地区：[1]沈阳化工大学信息工程学院

出　　处：《化工自动化及仪表》2023年第6期841-848,共8页Control and Instruments in Chemical Industry

摘　　要：为了进一步提高规范变量分析(CVA)辨识方法的准确性,降低在干扰因素下辨识精度的损失,提出改进CVA辨识方法。控制策略上比较两种经典的估计CVA型矩阵A和C方法,并分析这两种方法所获得的估计结果之间的差异。在此基础上,为消除误差,提出统一的估计系统矩阵A和C的方法,通过修正表达式的形式,将两种经典方法联系起来。仿真结果表明:辨识结果在保证精度的同时,得到的传递函数可以很好地拟合跟踪工业数据,而且辨识方法具有较好的鲁棒性。For purpose of further improving accuracy of CVA identification method and reducing loss of identification accuracy under interference factors,an improved CVA identification method was proposed.In terms of control strategy,two classical methods of estimating CVA matrix A and C were compared and differences between them were analyzed to propose a unified method for estimating system matrices A and C so as to eliminate the error there,and through modifying the expression,the two classical methods were linked.Simulation results show that,the transfer function obtained can well fit and track industrial data while ensuring the accuracy of identification results,and this identification method has good robustness.

关 键 词：改进CVA算法 系统辨识 状态空间模型 子空间辨识方法 拟合跟踪 工业数据 一阶误差 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]