摘要
针对Kalman滤波中存在量测野值的特点,依据滤波基本理论,提出了残差变化率概念。将残差变化率引入到野值判定标准中,根据量测值变化率的大小自动调整判断阀值,给出了判别野值的检验方法。由于采用了测量与估计的动态信息,使得野值的判定更为准确,并采用替代的方法对单个或连续野值加以修正,保证了Kalman滤波精确度与数据的连续性。仿真结果表明,当量测值出现较大变化时,这种方法能有效提高系统对野值的检测准确度和滤波精确度。
According to the character of the observation outliers and the basic filtcr theory, the conception of the variety of residual difference, which is introduced to the judgment of outlier and used to adjust the threshold automatically,is presented, and the test methed of distinguish outliers is given. Using the dynamic information of measurement and evaluation, the method makes the evaluation more accurate. Otherwise, substitute method is used to modify the single outlier or series outliers to ensure the precision of Kalman filter and the continuity of data. The simulation results show that the method can improre the accruacy of outlier detecting effectively.
出处
《电机与控制学报》
EI
CSCD
北大核心
2007年第2期188-190,195,共4页
Electric Machines and Control
基金
国家自然科学基金项目(60272051)
