采样数据含高频有色噪声的连续模型辨识

提出了一种采样数据中存在高频有色噪声的连续模型辨识方法,该方法通过引入积分运算,将连续时间系统的微分方程模型转换为积分方程模型,从而在辨识时使噪声的影响可以忽略,然后直接利用传统的最小二乘法估计出系统的连续时间模型参数。该方法具有较强的抗有色噪声干扰的鲁棒性、计算方法简单、参数辨识精度高等优点。仿真实例验证了该方法的有效性与可行性。

Identification of Continuous-time Model from sample data with high-frequency colored noise

A continuous-time model identification method from sample data with high-frequency colored noise is presented. By introducing an integral operation, the differential equation model of continuous-time system can be transformed into integral equation model, then the effect of the noise can be eliminated, and a continuous-time model parameter is estimated directly by using the conventional least square method. This method has the virtues of stronger robustness of anti-jamming for colored noise, simple and high precision of parameter identification. Simulation example demonstrates the effectiveness and the feasibility of the presented method.