具有不稳定初始状态的连续时间系统辨识

针对传统辨识方法不适用于具有不稳定初始状态的连续时间系统的问题,提出一种全新的状态估计辨识法.首先,用状态空间模型中状态变量的初始值表征系统初始状态,并将状态变量的初始值看作待辨识参数的一部分.然后,用粒子群优化算法获得所有参数的最优估计.该方法在测试开始前不需要任何过程数据,对测试信号无任何要求,可直接用于闭环辨识.仿真实验证明该算法是有效的.

Identification of continuous-time systems with unsteady initial conditions

A new state estimation identification method is proposed for the identification of the continuous-time systems with non-zero unsteady initial conditions, to which the traditional identification methods cannot be applied. Initial values of state variables representing the initial conditions of the systems are considered a part of the parameters to be estimated. The particle swarm optimization is then used to obtain the optimal estimations of all parameters. This method needs no process data before the test starts and has no requirement for the test signal. Moreover, it can be applied to closed-loop identification directly. Its effectiveness is demonstrated through simulations.