摘要
神经元动态规划是近年发展起来的一种优化方法 .它采用计算机仿真和函数近似 ,简化对状态空间的搜索 ,可以有效克服“维数危机” ,有广阔的应用前景 .本文对神经元动态规划作一综述 。
Neuro dynamic programming is an optimizaed method developed these years. Using the method of computer simulation and function approximation, it simplifies the search of state space and provides a effective way to overcome 'curse of dimensionality' so that it may be widely applied. This paper presents a brief over view of the research on neuro dynamic programming with a view to being helpful to the corresponding research work.
出处
《信息与控制》
CSCD
北大核心
2001年第4期343-347,351,共6页
Information and Control
基金
国家自然科学基金6 99740 39的支持
关键词
动态规划
神经元动态规划
计算机仿真
dynamic programming,neuro dynamic programming, approximation, simulation,temporal difference learning
