000003383 001__ 3383
000003383 005__ 20141118153401.0
000003383 04107 $$acze
000003383 046__ $$k2005-05-09
000003383 100__ $$aTurek, M.
000003383 24500 $$aSome notes on control of asynchronous electromotor by improved Carla method

000003383 24630 $$n11.$$pENGINEERING MECHANICS 2005
000003383 260__ $$bInstitute of Thermomechanics AS CR, v.v.i., Brno
000003383 506__ $$arestricted
000003383 520__ $$2eng$$aModifications of reinforcement learning algorithm, so called continuous action reinforcement learning automaton (CARLA), are presented in this contribution. Automaton learning algorithm is based on rewarding that gradually evolves the set of probability densities. This set is consequently used for action set determination. Modifications consist of improving learning parameters based on learned values. Thereby higher values of probability density near the best action are reached and therefore the variance of chosen actions is lower than original. The influence of modifications is proved by simulation study describing learning and behavior of asynchronous electromotor scalar control. Standard PSD controller is used whose parameter values represent actions of three independent automata. The goal of on line learning process is to minimize the mean square of control error. Here described modifications of algorithm allow the improvement of quality of revolutions control with preserving basic algorithm characteristics. 1 Introduction Despite the progress in development of control systems, the general problem of setting their parameters still remains unsolved. It is possible to calculate the parameters easily, when mathematical model of controlled system is known. In the other cases the analogy with similar system or experts practice are used. When this approach fails the appropriate method of artificial intelligence can be used. One of these methods is CARLA (e. g. [3]). Its function was successfully proved by practical applications (e. g. [2]) and results prove assumption that better results can be reached by CARLA method. The appropriate learning parameters [1] improve the behavior of CARLA method. The problem is that different characteristics are important in different phases of learning. First the high speed of learning, later the precision of learned value is needed. These two demands contradict each other when CARLA method is used. But both of them can be achieved by its modification. 1

000003383 540__ $$aText je chráněný podle autorského zákona č. 121/2000 Sb.
000003383 653__ $$a

000003383 7112_ $$aENGINEERING MECHANICS 2005$$cSvratka (CZ)$$d2005-05-09 / 2005-05-12$$gEM2005
000003383 720__ $$aTurek, M.$$iPulchart, J.$$iBřezina, T.
000003383 8560_ $$ffischerc@itam.cas.cz
000003383 8564_ $$s2564210$$uhttps://invenio.itam.cas.cz/record/3383/files/Turek-PT.pdf$$y
             Original version of the author's contribution as presented on CD, .
            
000003383 962__ $$r3238
000003383 980__ $$aPAPER