在现代测量数据误差多源性及数据处理高精度要求下，近三十年来 EIV（errors-in-variables）模型及整体最小二乘（total least squares, TLS）估计理论发展迅速。然而，以经典统计学为基础的TLS估计在处理现代空间测绘技术下含丰富时空先验信息数据或复杂分布数据等情况受限。本项目拟研究EIV模型的贝叶斯估计理论，从而将EIV模型的随机量信息扩展至最一般的情况，包括具有任意分布的观测数据和任意先验分布的参数；基于贝叶斯理论，推导EIV模型的最大似然估计、最大验后估计和贝叶斯估计三种严密算法；采用数理统计理论以及蒙特卡洛算法，研究贝叶斯估计的统计特性，包括参数的估计精度、单位权方差以及方差分量估计。EIV模型的贝叶斯估计提供了复杂分布下的严密估计方法，并利用参数的先验信息提高估计的可靠性，研究成果对于完善EIV模型估计理论具有重要理论意义和应用价值。

As geodetic observation techniques have varied rapidly, the Errors-In-Variables (EIV) model and Total Least-Squares have been intensively investigated quite recently. However, the investigation of the EIV model is still limited for the geospatial data analysis, because the corresponding adjustment theory is still very limited due to the prerequisites of the stochastic model. In this project, we propose the Bayesian inference of the EIV model in terms of the given probability density function, which includes the maximum likelihood estimation, maximum a posteriori estimation and the beyesian estimation of unknown parameters. In addition, the traditional statistical method and the Monte Carlo strategy is proposed to analytically provide and numerically compute the statistical analysis of the parameter estimates including the variance factor, dispersion matrix and variance covariance components. The framework of the Beyesian statistics within the EIV model, which considers the prior information, could provide a more reliable results and ought to coincide with real applications, which makes progress for certain fundamental geodetic problems from a theoretical point of view.