基于渐近平稳过程的修正扩展卡尔曼滤波算法

An Improved Extend Kalman Filtering Algorithm Based on Asymptotic Stationary Process

多目标跟踪中由于RCS闪烁和杂波变化造成的测量噪声波动,导致测量噪声并不是完全服从高斯分布。研究了基于渐近平稳过程的噪声模型,引入测量值置信度函数,提出了一种将测量值置信度反馈至滤波过程的方法,通过修正新息协方差阵及增益矩阵,降低测量噪声波动对滤波的影响。该方法增强了噪声压缩能力,有效地抑制了滤波的误差尖峰。最后,仿真结果验证了所提方法的有效性与可行性。

In multitarget trackingthe measurement noise is not an absolute Gaussian distribution because of clutter and the undulation of RCS.A model of noise was studied based on asymptotic stationary processand a confidence function was introduced.A method that could feed back the confidence degree of the measured value to the filtering process was proposed.By correcting the gain matrix and covariance matrixthe effect of measuring noise fluctuation on filtering was reduced.The method could enhance the ability of the noise compression and restrain the peak error.Simulation result verifies the validity and feasibility of the method.