IEEE Transactions on Automatic Control · 1979 · 122 citations · 6 references
EngineeringMathematicsDynamical SystemsProbability TheoryStochastic ProcessesAdaptive ControlControl SystemsControl TheoryMarkov ChainsMarkov ProcessesStochastic ControlFinite Markov ChainsState Transition ProbabilitiesState-dependent Transition ProbabilitiesTime-varying Transition ProbabilitiesTransition Probabilities
Consider a controlled Markov chain whose transition probabilities depend upon an unknown parameter α taking values in finite set <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">A</tex> . To each α is associated a prespecified stationary control law <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">\phi(\alpha)</tex> . The adaptive control law selects at each time <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">t</tex> the control action indicated by <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">\phi(\alpha_{t})</tex> where α <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">t</inf> is the maximum likelihood estimate of α. It is shown that α <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">t</inf> converges to a parameter α <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">*</sup> such that the "closed-loop" transition probabilities corresponding to α <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">*</sup> and <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">\phi(\alpha^{\ast})</tex> are the same as those corresponding to α <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">0</sup> and <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">\phi(\alpha)</tex> where α <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">0</sup> is the true parameter. The situation when α <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">0</sup> does not belong to the model set <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">A</tex> is briefly discussed.
6
Karl Johan Åström, Björn Wittenmark · Automatica · 1973 · 1.6K citations
Estimation and control in Markov chains
Petr Mandl · Advances in Applied Probability · 1974 · 180 citations
Bayesian Decision Problems and Markov Chains
Paul H. Randolph, J. J. Martin · Technometrics · 1968 · 90 citations