针对平台误差系数建模预测问题，提出了基于多变量相空间重构的多输出高斯过程回归预测算法。通过多变量相空间重构将两个相关性较强的平台误差系数重构在一个相空间中，采用多输出高斯过程回归模型同时预测这两个平台误差系数。该算法充分利用了两个误差系数之间的相关性，提高了预测精度，而且可以得到任意置信度下的预测均值和置信区间，为解决平台误差系数建模预测提供一条新的途径。

A MultipleOutput Gaussian Process Regression (MOGPR) forecasting algorithm based on multivariate phase space reconstruction (MPSR) was proposed to solve the problem of platform error coefficient modeling and forecasting.Two tensely correlated platform error coefficients were reconstructed in one phase space based on MPSR and they are forecasted together with the MOGPR.The correlation of the two error coefficients was made full use of in this algorithm to improve the forecasting accuracy.The forecasting average and range of confidence under any degree of confidence could be obtained.It provides a new method for solving the problem of platform error coefficient modeling and forecasting.