论文部分内容阅读
本文应用预报线性化方法研究了大系统在线参数辨识问题。通过对大系统进行分级分解处理,将系统分为若干互相联系的子系统,分二级迭代地估计状态变量和系统参数。第一级用不变式嵌入法求解优化的二点边值问题,给出子系统的最优解;第二级应用预报插值法求得协调变量。这种方法不仅简化了上机前的公式推导,而且提高了计算速度,并给出满意的结果。为实时辨识大系统创造了条件,成为大系统辨识行之有效的手段。
In this paper, the problem of on-line parameter identification of large-scale system is studied by means of forecasting linearization. By hierarchically decomposing the large system, the system is divided into several interconnected subsystems, and the state variables and system parameters are estimated iteratively in two steps. In the first stage, the invariant embedded method is used to solve the optimal two-point boundary value problem, and the optimal solution of the subsystem is given. The second level is obtained by the interpolation method of predictive variables. This method not only simplifies the pre-machine formula derivation, but also improves the calculation speed and gives satisfactory results. It has created conditions for real-time identification of large systems and has become an effective means of identifying large systems.