AN ADAPTIVE AUTOMATON CONTROLLER FOR DISCRETE-TIME MARKOV PROCESSES

被引:19
作者
RIORDON, JS
机构
[1] Faculty of Engineering Carleton University, Ottawa
关键词
D O I
10.1016/0005-1098(69)90085-5
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
The computation of an optimal feedback controller characteristic for a non-linear stochastic system may be facilitated by the use of a stochastic automaton as a system model. A problem of particular interest is that of a long duration stationary Markov process in which the state is observable but the process dynamics and disturbance characteristics are initially unknown. The determination of a suitable control algorithm, in the form of an adaptive automation in the feedback loop, is considered in this paper for such a process. Since the algorithm is to be used on-line to perform simultaneously the functions of estimation and control, it must constitute an efficient convergent multi-stage dual control strategy. It is shown that an existing method for dual control of a repetitive single-stage stochastic process may be extended to apply to the present case. A method is introduced of calculating successive policy estimates recursively, so that the task of updating the estimated optimal feedback policy at each stage of the process is rendered feasible. The application of the automaton controller is illustrated by the simulated adaptive control of a non-linear conditionally stable heat treatment process disturbed by multiplicative noise. © 1969.
引用
收藏
页码:721 / +
页数:1
相关论文
共 22 条