在加性噪声背景下,将线性系统辨识问题纳入到独立分量分析(ICA)瞬时混合模型中,提出了一种新的结合ICA技术的线性系统辨识算法。该算法无需对噪声信号的统计特性进行任何先验假定,根据ICA模型中瞬时混合矩阵的先验知识,准确估计出系统输出的加性噪声,从观测数据减去该噪声后,利用最小二乘理论实现算法对未知系统的参数估计。仿真实验表明,该算法不仅有效地消除了一般算法对于噪声特性的限制,而且在低信噪比下仍能取得较为理想的辨识结果。 In the context of additive noise, the linear system identification problem is integrated into the ICA transient mixed model, and a new linear system identification algorithm based on ICA is proposed. According to the prior knowledge of instantaneous mixing matrix in ICA model, this algorithm can accurately estimate the additive noise of the system output. After subtracting the noise from the observed data, the algorithm uses the least square The Parameter Estimation of Unknown System by Algorithm of Theoretical Realization. The simulation results show that this algorithm not only effectively eliminates the limitation of the general algorithm for the noise characteristics, but also achieves a better identification result under the low SNR.