|
| |
|
| |
|
|
Determination of Q-function optimum grid applied on active magnetic bearing control task
Březina, Tomáš ; Krejsa, Jiří
AMB control task can be solved using reinforcement learning based method called Q learning. However there are certain issues remaining to solve, mainly the convergence speed. Two-phase Q learning can be used to speed up the learning process. When table is used as Q function approximation the learning speed and precision of found controllers depend highly on the Q function table grid. The paper is denoted to determination of optimum grid with respect to the properties of controllers found by given method.
|
|
|
Rotor supported on magnetic bearings - numerical and experimental analysis
Půst, Ladislav ; Kozánek, Jan ; Veselý, Jan
The dynamic properties of rotor supported on passive magnetic bearingsis investigated by means of numerical solution and by measurement on the prototypeof experimental set developed in Institute of Thermomechanics AS CR. Mathematicalmodel was constructed both for linearized part of rotor with bearing and for stronglynonlinear system including impacts in retainer bearings. Comparison with resultsof measurement on prototype enables to ascertain real parameters of linearized partof magnetic bearings.
|
|
|
Operation testing of the rotor supported on aerodynamic bearings
Vaník, J. ; Šafr, M. ; Šimek, J. ; Kozánek, Jan ; Veselý, Jan
A high-speed synchronous electric machine has been modified to run on contactless aerodynamic bearings designed and manufactured within this project. The experience gained during putting the motor in operation at the Institute of Thermodynamics is described and the first results from the test runs at speeds up to 50,000 rpm (basic diagnostic measurement of rotor relative vibration and subsequent spectral identification) are presented.
|
|
|
Pseudospectra and non-stationary excitation
Kozánek, Jan ; Berthelot, F.
The behaviour of linear (and non-linear) dynamical systems is characterized with the help of the spectral properties. The pseudospectrum is in this situation better instrument, giving the estimation about the resolvent. The formal substitution (using a regression method) of the non-stationary excitation of rotating machines during start up and run down regimes by the exponencial function corresponding to the resolvent solution is proposed and discussed.
|
|
| |
|
| |
|
| |
|
|
Stochastic policy in Q-lerning used for control of AMB
Březina, Tomáš ; Krejsa, Jiří ; Věchet, S.
A great intention is lately focused on Reinforcement Learning (RL) methods. The article is focused on improving model free RL method known as Q-learning used on active magnetic bearing model. Stochastic strategy and adaptive integration step increased the speed of learning approximately hundred times. Impossibility of using proposed improvement online is the only drawback, however it might be used for pretraining and further fined online.
|
guest ::
