中文摘要：

利用滤波正常时的残差向量必须服从正态分布的性质,提出1种滤波是否异常的假设检验方法,判断动态模型误差和观测粗差对变形监测滤波解的影响。研究结果表明：当滤波出现异常时,通过给定置信度的假设检验,确定自适应因子来调节误差较大的预报信息,从而较好地消除模型误差的影响;利用自适应抗差滤波解决变形监测数据处理中整体平衡观测信息和预报信息的贡献的问题,并通过实例验证算法的有效性和优良性。

英文摘要：

Based on the statistical property that residual vectors should be normally distributed, a hypothesis test was proposed whether there is an exception in filter to control the influence of dynamic model errors and great error in observation on the status parameter estimation. The results show that when there is an exception in filter, an adaptive factor can be determined by means of a hypothesis test with a given confidence level, which can adjust forecast information with big error and resist the influence of modeling error. A robustly adaptive Kalman filter algorithm was given to solve the overall balance contribution between observations and forecasts in the deformation monitoring data processing. Then, the validity and excellent of the algorithm are verified through a example of deformation monitoring.