带有乘性噪声的时滞系统多传感器信息融合

研究了带有乘性噪声的线性时滞系统的局部最优预报估计和全局最优线性加权信息融合问题。 通过虚拟噪声补偿技术,将该问题转化为一类带有未知时变噪声的随机系统的状态最优估计问题。基于等价系统的新息重组分析及Hilbert空间上的投影定理,给出局部最优预报器设计,进而通过求解与各单传感器子系统有相同维数的Riccati方程得到多传感器分布式全局最优加权信息融合算法。与集中式融合估计算法相比,该方法无需扩维。最后通过一个仿真实例证明该算法的有效性。

Multi-sensor optimal information fusion for time-delay systems with multiplicative noise

The locally optimal predictor and globally optimal linear weighted fusion algorithm were investigated for linear time-delay systems with multiplicative noise. The problem could be transformed into an optimal state estimation problem of a stochastic system with unknown time-varying noises through compensation of fictitious noises. The locally optimal predictor was presented by the reorganized innovation approach and projection theory in Hilbert space, and then the globally optimal linear weighted fusion algorithm was given by solving Riccati-type equations with the same dimension as the sensor subsystems. Compared with the centralized optimal filter by augmentation, the proposed approach could reduce the complex of calculation. A numerical example and its simulation results were given to show the effectiveness of the proposed method.