论文部分内容阅读
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。
To solve the problem of estimation in large-scale networked systems with bounded noises, a distributed algorithm based on collaboration is proposed. Consider the interaction between the various subsystems, the establishment of a completely decoupled model and the interaction model composed of synthetic model. According to the existence of any bounded noise in the system, the objective function of each subsystem constitutes a convex combination of the system-wide objective function to make full use of the information of each system. The algorithm is simple to model, and the subsystems adopt parallel computing, so the computational efficiency is high. When applied to the model predictive control, the problem that the state is not completely measurable is solved. The simulation results show that the estimation accuracy of the collaborative distributed estimation algorithm is higher than Kalman estimation algorithm, and the estimation error is less than 0.05.