热工模型中纯迟延时间的递归估计算法

Recursive Estimation Algorithm for Calculating Time Delay in Thermodynamic Process

为了对典型的热工模型——二阶惯性加纯迟延模型的参数进行在线辨识,介绍并推导了基于递推最小二乘在线辨识算法,对带有纯迟延时间系统进行参数估计。连续系统小于采样周期的纯迟延时间,在相应的采样系统中就会产生一个副实零点,利用这个零点引起的相位变化来对小于采样周期的迟延进行估计。采用最小化损失函数的两步辨识法,对系统的其它参数和采样周期整数倍的迟延进行估计。利用这种递推辨识算法对热工模型进行参数辨识,得到模型的离散传递函数,进而求得其对应的连续传递函数。结果表明,这种算法收敛性好,能够用来对热工对象进行在线参数估计。

In order to estimate thermodynamic process which is a model of second-order inertia transfer function plus a time delay, this paper introduces an on-line estimation method based on standard recursive least squares for the estimation of both time delay and other parameters. The algorithm relies on the fact that if the continuous time model has a delay or anticipation shorter than one sampling time, then a real negative zero arises in the corresponding sampled system. By inspection of the phase contribution of this zero, the value of the delay is recursively updated. Other parameters and time delay which are integer times of sampling time can be obtained by minimizing the cost function in a two-step estimation algorithm. Using this theory the discrete parameters of thermodynamic process can be obtained and further the continuous parameters. Simulation results prove the convergency of the time delay estimation algorithm used in on-line estimation for thermodynamic process.