针对机动目标的改进UFIR跟踪算法

在机动目标跟踪中,卡尔曼滤波器(KF)及其改进算法的性能依赖于过程噪声统计特性的准确性,若模型过程噪声与实际存在偏差,通常会出现估计误差增大甚至发散的现象.无偏有限冲击响应滤波器(UFIR)在滤波过程中无需过程噪声统计特性的先验知识,将其应用于机动目标跟踪中,针对现有UFIR滤波器中广义噪声功率增益(GNPG)不随量测新息变化的问题,设计了一种根据相邻时刻量测新息比值动态调整GNPG的改进UFIR滤波器,改善了UFIR滤波器的机动检测能力.仿真结果表明,当假定过程噪声准确时,现有和改进UFIR滤波器与KF的跟踪性能相似;但当假定过程噪声不准确时,改进UFIR滤波器具有最佳的滤波效果.

Maneuvering target tracking with modified unbiased FIR filter

In the field of maneuvering target tracking, the performance of Kalman filter(KF)and its variants is dependeds on the accuracy of the assumed process noise statistics. If the assumed process noise is not accurate, the performance of the KF and its improved algorithms will be degraded significantly. In some cases, the filters might even cannot be converged. Unbiased finite impulse response (UFIR) filter does not need the prior knowledge of the process noise statistics during filtering. Hence, it can be utilized to overcome the problem of the inaccurate assumed process noise statistics to realize the maneuvering target tracking. Since the generalized noise power gain (GNPG) of the existing UFIR filter cannot be adapted to the measurements innovation, an improved UFIR filter was proposed. The proposed UFIR dynamically adjusts GNPG according to the ratio of measurements innovations between the adjacent time such that it can improve the detecting ability of the UFIR filter for target maneuver. The simulation results illustrate that if assumed process noise is accurate, the performance of the existing UFIR filter and the proposed FIR filter is similar to KF; but if assumed process noise is not accurate, the performance of the proposed UFIR shows better than the other ones.