当以M序列信号作为输入信号时，传统的相关辨识方法在序列长度小的情况下辨识精度不高且容易受到噪声干扰的影响。在研究先前相关辨识法的基础上，利用带有遗忘因子的最小二乘方法对相关辨识法进行了改进。仿真结果表明，在不缩短单个序列时间间隔Δt的条件下，改进的方法有着更高的辨识精度和更强的抗干扰能力，从而降低了对测试系统硬件设计的要求。因此，改进的相关辨识法在参数辨识领域的工程应用上有着广阔的应用前景。

When M-sequence signal is used as actuating signal,the system identification accuracy of traditional correlation method is not high enough and is liable to be influenced by system noise as the length of M-sequence is short.On the basis of study of traditional correlation identification,the authors proposed a modified correlation identification method based on the least square with forgetting factor.Simulation results indicated that the modified method has higher accuracy and anti-interference capability without cutting down the single sequence interval (Δt),thus can reduce the requirements to system hardware.The modified method may find a wide application in parameter identification.