文章基本信息

标题：Approximate Dynamic Programming via Penalty Functions * * This research was partially funded by the European Commission under the project Local4Global.
本地全文：下载
作者：Paul N. Beuchat ; John Lygeros
期刊名称：IFAC PapersOnLine
印刷版ISSN：2405-8963
出版年度：2017
卷号：50
期号：1
页码：11814-11821
DOI：10.1016/j.ifacol.2017.08.1993
语种：English
出版社：Elsevier
摘要：AbstractIn this paper, we propose a novel formulation for encoding state constraints into the Linear Programming approach to Approximate Dynamic Programming via the use of penalty functions. To maintain tractability of the resulting optimization problem that needs to be solved, we suggest a penalty function that is constructed as a point-wise maximum taken over a family of low-order polynomials. Once the penalty functions are designed, no additional approximations are introduced by the proposed formulation. The effectiveness and numerical stability of the formulation is demonstrated through examples.
关键词：KeywordsStochastic optimal control problemsApproximate dynamic programmingSoft constraints