Fully probabilistic control design in an adaptive critic framework

0364820 - ÚTIA 2012 RIV GB eng J - Journal Article
Herzallah, R. - Kárný, Miroslav
Fully probabilistic control design in an adaptive critic framework.
Neural Networks. Roč. 24, č. 10 (2011), s. 1128-1135. ISSN 0893-6080. E-ISSN 1879-2782
R&D Projects: GA ČR GA102/08/0567
Institutional research plan: CEZ:AV0Z10750506
Keywords : Stochastic control design * Fully probabilistic design * Adaptive control * Adaptive critic
Subject RIV: BC - Control Systems Theory
Impact factor: 2.182, year: 2011
http://library.utia.cas.cz/separaty/2011/AS/karny-0364820.pdf

Optimal stochastic controller pushes the closed-loop behavior as close as possible to the desired one. The fully probabilistic design (FPD) uses probabilistic description of the desired closed loop and minimizes Kullback–Leibler divergence of the closed-loop description to the desired one. Practical exploitation of the fully probabilistic design control theory continues to be hindered by the computational complexities involved in numerically solving the associated stochastic dynamic programming problem; in particular, very hard multivariate integration and an approximate interpolation of the involved multivariate functions. This paper proposes a new fully probabilistic control algorithm that uses the adaptive critic methods to circumvent the need for explicitly evaluating the optimal value function, thereby dramatically reducing computational requirements. This is a main contribution of this paper.
Permanent Link: http://hdl.handle.net/11104/0200201