在基于Vasicek利率期限结构模型,使用Kalman滤波器进行状态估计时,通常采用量测噪声不相关假设。研究表明,量测噪声在很多情况下满足相关性假设。本文采用一般性的相关量测噪声假设,提出针对Vasicek模型潜变量估计的状态扩维Kalman滤波法。实验结果表明,量测噪声相关假设下的Vasicek模型比较准确地描述了利率期限结构的动态变化特征。 Based on Vasicek’s term structure model, the Kalman filter is used to estimate the state, and the noise-free assumption of measurement is usually adopted. Research shows that measurement noise satisfies the correlation hypothesis in many cases. In this paper, a general state-of-the-art measurement noise assumption is proposed to propose a state-expanded Kalman filter for the estimation of latent variables in Vasicek model. The experimental results show that the Vasicek model under the assumption of measurement noise accurately describes the dynamic characteristics of the term structure of interest rates.