连续线性系统模型参数直接辨识中噪声模型处理方法

Method of noise treatment model in direct parameters identification of continuous linear system model

针对线性连续系统,利用hermite调制函数,通过对连续模型的数字调制积分得到直接以连续模型参数表示的离散等价辨识模型,在此基础上引入一种噪声模型,来替代表示输入和输出端的噪声.当系统含有有色噪声,用普通最小二乘法辨识无法得到较准确的辨识参数时,采用引入噪声模型的方法得到的辨识参数估计值仍有较高的精度.仿真算例表明了该方法具有一定的优越性.

For linear continuous-time systems using hermite modulation function,through the digital modulating integral to the continuous-time model,an equivalent discrete identification model with continuous time model parameters is developed.Based on the introduction of a noise model,input and output side noise are replaced.When the system contains a colored noise,using ordinary least square identification,be obtained more precise recognition results could not,the introduction of noise model parameter estimation methods were still with high recognition accuracy.Two examples were given to illustrate this method.