论文部分内容阅读
交互式多模型滤波(IMM)的交互环节使得系统状态量不再服从单纯的高斯分布,用现有方法对其概率分布的估计存在较大的误差.对此,考虑到模型的混合概率是时变的,IMM的交互过程可以用非线性方程来描述,因而采用容积卡尔曼滤波(CKF)中的容积法则对高斯随机变量经非线性函数传播后的概率分布进行估计,并从理论上证明了容积法则的近似精度.仿真实验表明,由于提高了对交互后随机变量概率分布的估计精度,所提出的方法能够有效改善IMM在量测噪声较大时的滤波效果.
The interaction of IMM makes the system state no longer obey the simple Gaussian distribution, and there is a big error in the estimation of its probability distribution by the existing methods. In this regard, considering the probability that the model’s mixture probability is The IMM interaction process can be described by nonlinear equations. Therefore, the volumetric law of volumetric Kalman filter (CKF) is used to estimate the probability distribution of Gaussian random variables after they are propagated by non-linear function, and it is theoretically proved that The approximate accuracy of the volumetric rule is verified by the simulation results. The proposed method can effectively improve the filtering performance of the IMM when the measured noise is large due to the improved estimation accuracy of the probability distribution of the random variables after the interaction.