观测信息一步时滞分享模式LQG系统的分散反馈主从对策研究

ON DECENTRALIZED FEEDBACK STACKELBERG GAME OF LQG SYSTEMS WITH ONE-STEP DELAY INFORMATION SHARING PATTERN

考察了具有观测信息一步时滞分享模式(One Step Delay Information Sharing Pattern,简记为1-SDISP模式)系统的分散信息估计和反馈主从对策(正则型)问题。对随机动态系统内各决策者之间观测信息的分享具有一步时滞时,提出了'反状态滤波'概念,并导出了该信息模式下的状态滤波和反状态滤波的递推估计公式。以此为工具,获得了1-SDISP模式LOG反馈主从对策(正则型)问题的最优策略解。解的形式表明它关于决策者对系统内所有决策者最优滤波的估计成线性关系。

In this paper, for linear—quadratic—Gaussian (LQG) systems with one—step delay information sharing pattern (i—SDISP), the problems of information estimator and feedback stackelberg game are considered. The formulae of optimal information estimator are derived. The solution of the LQG feedback Stackelberg game with i—SDISP is obtained. It is shown that the solution is linear in the decision maker's optimal estimations of all of the decision maker's optimal state filterings.