基于协作的分布式估计算法及在模型预测控制中的应用

Distributed estimation algorithm based on cooperation for model predictive control

为解决带有有界噪声的大型网络化系统中的估计问题,提出一种基于协作的分布式估计算法。考虑各个子系统间的相互作用,建立一个由完全解耦模型和相互作用模型组成的合成模型。根据系统中存在的任意有界噪声,由各个子系统目标函数的凸组合组成全系统的估计目标函数,来充分利用各个系统的信息。该算法建模简单,各子系统采用并行运算,计算效率高;应用到模型预测控制中,解决了状态不完全可测的问题。仿真结果表明基于协作的分布式估计算法的估计精度比Kalman估计算法更高,估计误差小于0.05。

A distributed estimation algorithm based on cooperation was developed for estimates in large networked systems with bounded noise.Composite models combined of complete decoupling models and interaction models were developed based on the interaction among subsystems.A system-wide objective function was defined based on a strong convex combination of all the subsystems objective functions for an arbitrary bound of the system noise to make full use of the information in all the subsystems.The model is easy to...