飞行试验中，异常值是影响光电经纬仪激光测距精度的主要问题。能否准确地剔除异常值，成为试验结论是否正确的首要条件。通过分析无迹卡尔曼滤波UKF和交互多模型IMM的原理及数学模型，提出了可应用于光电经纬仪实时数据处理的改进算法--交互多模型的无迹卡尔曼滤波算法(IMM-UKF)，在此基础上给出了算法数学模型，最后通过飞行试验验证了该算法的工程实用性。试验结果表明，该算法有效去除了激光距离中的异常值，提高了估计精度及算法稳定度。

In the flight test,outlier is the main factor that has effect on laser range-finding of the photoelectric theodolite.Eliminating the outliers accurately has great importance for obtaining the correct test result.We analyzed principle of Unscented Kalman Filter (UKF) and Interacting Multiple Models (IMM),proposed an improved algorithm,IMM-UKF,for real-time data processing of photoelectric theodolite.On which basis,we gave the mathematical model of the algorithm,and proved the validity through flight test.The test results indicated that this algorithm can eliminate the outliers of laser effectively,and improve the estimating precision and stability.