摘要
通过分析系统模型误差和预报残差对卡尔曼滤波的影响以及区间矩阵对滤波效果的影响,提出了在基于方差分量的自适应滤波器的基础上添加置信区间的方法来提高预测精度。一方面,自适应滤波器可以通过方差分量不断地校正伪观测值,克服了由于动态目标的机动性引起滤波发散的缺陷;另一方面,通过预测误差添加置信区间不仅可以修正状态,还能提供状态预测的范围。仿真结果表明,该算法的滤波效果优于基于传统的卡尔曼滤波算法的状态估计,适用于机动目标定位数据的实时处理。
By studying the influence of system state model error and the prediction error on the Kalman filter and the influence of the interval matrix on the filtering effect,a new method to improve the prediction accuracy is put forward by the addition of confidence interval on adapative Kalman filter. On the one hand, adaptive filtering can overcome the filtering divergence caused by dynamic target maneuvering. On the other hand,the confidence interval of prediction error can be used to modify the predicted state and provide a range of state prediction. Simulation results show that the filtering effect is better than the traditional Kalman filter,and can be used for real-time data processing of dynamic positioning.
出处
《北京信息科技大学学报（自然科学版）》
2017年第6期24-28,33,共6页
Journal of Beijing Information Science and Technology University
基金
国家自然科学基金资助项目(61271198)
北京市科技提升计划项目(5211624101)
关键词
卡尔曼滤波
自适应
置信区间
状态估计
Kalman filer
adapative filter
confidence interval
state estimation
