ALGORITHM OF ADAPTIVE OUTPUT CONTROL OF LINEAR SYSTEM WITH IMPROVED PARAMETRIC CONVERGENCE

Abstract

The problem of performance improvement of adaptive output control for linear time-invariant plant is addressed in the paper. The parameters of the plant are assumed to be unknown. A new method based on dynamical extension of regressive error model is proposed for design of the algorithms of controller parameters tuning. With a view of the proposed approach development the existing solutions to this problem are considered and analyzed. The first solution is based on gradient adaptation algorithm that drives control error to zero, however with arbitrary slow rate of convergence. The second solution uses identification algorithm of dynamical regressor extension. The algorithm has potentially high rate of convergence, however does not ensure decaying of control error. Thus, by modification of aforementioned solutions, namely, by dynamical extension of augmented error model and gradient adaptation algorithm the new algorithm is obtained. The algorithm permitsto increase the rate of the tuning of controller parameters arbitrary and drives control error to zero. The important property of the algorithm is that it does not require parameters identification neither of plant nor of its parameterized representation. This property relaxes the dependence of the control algorithms from persistent excitation condition, that is, the key identifiability condition. The proposed solution is verified in MatLab/Simulink environment by comparison with solution based on the gradient adaptation algorithm.

Keywords

• адаптивное управление с эталонной моделью
• алгоритм адаптации с улучшенной сходимостью