多传感器矩阵加权信息融合预测控制算法  

A predictive control algorithm for multi-sensor information fusion weighted by matrices

作　　者：李云[1] 邢宗新[2] 李世军 金浩[1] 赵明[1] 张玉茹[1] 

机构地区：[1]哈尔滨商业大学计算机与信息学院,黑龙江哈尔滨150028 [2]哈尔滨商业大学科技处,黑龙江哈尔滨150028 [3]齐齐哈尔市邮政局计算机中心,黑龙江齐齐哈尔161000

出　　处：《应用科技》2012年第2期36-40,共5页Applied Science and Technology

基　　金：黑龙江省自然科学基金资助项目(F201015)

摘　　要：针对多传感器线性离散时不变随机控制系统,应用Kalmam滤波方法,基于状态空间模型,在线性最小方差最优信息融合准则下,给出了多传感器按矩阵加权信息融合预测控制算法.该算法将信息融合Kalman滤波器和预测控制相结合,避免了求解复杂的Diophantine方程,可明显减轻计算负担.与单传感器情形相比,可显著提高控制精度.一个三传感器目标跟踪控制系统的仿真例子说明了算法的有效性和正确性.Aiming at the multi-sensor discrete-time linear time-invariant stochastic controllable system based on state space model, a predictive controt^al^Crithm for multi-sensor information fusion weighted by matrices is presented using the Kalman filtering method under the linear minimum variance optimal information fusion criterion. This algorithm combines the information fusion of Kalman filter with predictive control, avoiding to solve the complex Diophantine equation, and therefore, it can reduce the computational burden obviously. Comparing to the case of a single sensor, the accuracy of the predictive control has been evidently improved. A simulation example of a 3-sensor target-tracking controllable system shows that it is effective and correct.

关 键 词：预测控制 信息融合 状态空间模型 矩阵加权 

分 类 号：O211.64[理学—概率论与数理统计]