利用混合卡尔曼滤波的混沌信号盲估计

Chaotic Signals Blind Estimation Based on Mixture Kalman Filter

针对混沌动力学系统时变参数未知的混沌信号,在含有状态噪声的情况下,利用混合卡尔曼滤波提出一种盲估计算法.对未知参数和混沌状态构成的高维状态进行估计,先利用卡尔曼滤波给出线性高斯部分的最优精确估计,剩余部分利用粒子滤波方法给出次优估计,文中详细研究了高斯噪声以及非高斯噪声下的最优重要性函数选取并推导了重要性权重的计算公式,最终基于有效粒子的最小均方误差准则实现了信号的盲估计.仿真结果表明该算法能有效实现含有状态噪声混沌信号的盲估计,并取得了比基本粒子滤波算法更优的性能.

Considering the state noise,a novel blind estimation algorithm for the chaotic signals with unknown time-varying parameters is proposed based on MKF（Mixture Kalman Filter）.The signals estimation is achieved by estimating the joint posterior probability density of the joint states.The linear Gaussian part of the joint states is estimated by Kalman filter,while the remain is estimated based on Particle Filter.The selection of the importance function is discussed in detail while the noise is Gaussian or not,and the importance weight formula is deduced.The states are estimated based on MMSE of effective particles finally.The simulation results show that the proposed method can solve the blind estimation of the chaotic signals with state noise effectively,and performance is better than the general method.