新型抗野值的Kalman滤波器研究

RESEARCH ON NEW KALMAN FILTER RESTRAINING OUTLIERS

在Kalman滤波的量测值中,经常出现连续、成片的野值,不仅严重地影响了其稳定性,而且造成滤波精度下降。针对这一问题,提出一种基于可调因子的新型抗野值Kalman滤波器方法。该方法以野值新息均值和均方差为输入变量,利用模糊控制器改变可调因子的大小,从而实现新息的自动调整,即当新息较小时,对野值进行修正,而当新息较大时,剔除野值。仿真结果显示,该方法抗野值效果良好,提高了滤波精度。

In measurements data of Kalman filtering, frequently occurred continuous and dense outliers not only seriously affect its stability but also make the precision of the filtering descended. To solve this problem, we propose a novel adjustable factor-based anti-outlier method for Kalman filter. The method uses fuzzy controller to alter the adjustable factor by taking the average value and mean square deviation of outliers＇ innovation as the controller＇ s fuzzy input variables, thus the automatic modification of the innovation is realised, that means, the outliers are to be modified when there is minor innovation but to be eliminated then there are more innovations. Simulation results prove that the method has perfect effect in restraining the outliers and high precision in filtering.