摘要
关联大系统的特点是维数高、内部子过程间相互关联,使得辨识方法的计算量和存储量急剧增加以及它本身的复杂性,以致常规辨识方法难以实现。为了减少大系统辨识的计算量,避免本身所带来的辨识困难,提出了获得其可分稳态模型的强一致性估计的分散辨识方法。该方法仅使用设定点的阶跃信号作输入辨识信号,并且每个子过程的输入输出和稳态模型的辨识都是在相应的局部单元完成的,因而大大减少了对过程的干扰和信息的交换量,该方法简单易懂,仿真结果说明了该辨识方法的有效性和实用性。
In order to reduce the computation in system identification,and make it easy,a scattered identification method with strong consistency in obtaining dividable steady-state model was proposed.This simple method took only set-point＇s step input signal as identification signal,and identified each sub-process input/output and steady-state model within relevant local units,thus disruption of the process and information exchange volume could be reduced.The simulation results show both validity and high accuracy of this identification method.
出处
《化工自动化及仪表》
CAS
北大核心
2010年第4期22-25,共4页
Control and Instruments in Chemical Industry
关键词
关联大系统
强一致性估计
阶跃信号
分散辨识
large-scale interconnected systems
strong consistency of estimates
step signal
scattered identification
