未知初始条件下连续时滞二阶系统的识别

在实际工业生产中有很多被控对象或控制过程都可被视为二阶纯滞后SOPDT(Second Order Plus Delay Time)的形式。在同时含有负载干扰及未知非零初始条件的情况下,针对待辨识参数过多而导致一些现有方法在实施最小二乘后只能获得部分参数的问题,提出首先给系统施加两次同方向的阶跃信号并对输出采样,进而对系统微分方程通式两边进行二重积分以获得两组回归方程,最后联立两次最小二乘结果便可获得所有待估计的参数值。

IDENTIFICATION OF SECOND-ORDER SYSTEM WITH CONTINUOUS TIME-DELAYS IN UNKNOWN INITIAL CONDITION

In practical industry production,quite a few controlled objects or controlling processes can be regarded as in forms of SOPDT(second-order plus delay time).In circumstance of with both the load disturbance and non-zero and unknown initial conditions,and in light of the problem that too many unknown identification parameters lead to some of the existing methods can only get partial parameters after the least-squares method is adopted,we propose that firstly,the step signals with same direction are injected into system twice and the outputs are sampled,then the double integral is operated on both sides of the system general differential equation to obtain two groups of regression equations,at last,all the parameters to be estimated can be acquired by combining the two results of least-squares method.