利用工业大系统动态信息建立稳态模型及其强一致性分析

Steady-State Establishment of the Large-Scale Industrial Process by Use of the Dynamic Information and Its Strong Consistency Analysis

本文提出了一种在相当弱的条件下,仅仅利用优化过程中系统设定点例行阶跃变化作为激励信号,对各子系统并行使用简单最小二乘法和近似动态线性模型,充分运用大系统的动态信息,得到了大系统稳态模型的一致估计的理论证明,并利用数字仿真进一步验证了其方法的有效性,还给出了一种实用的,能强一致估计线性渐近定常大系统稳态模型的方法。

In this paper, under the weak assumptions such as only the step changes of system set points being the exciting signal, unknown structures and the dynamic parameters, approximate linear dynamic model and the least squares estimation method, for the linear slow time-varying large-scale system, the steady-state gain estimate is formed from the estimates of the parameters and a parallel identification algorithm is put forward. The consistency of the estimate and the convergence of the parallel iteration are analyzed. Based on this consistency theorem, a pragmatic method to get the strong consistent estimate of the steady-state model of the large-scale system is presented. Simulation study has already proved this point.