A nearly optimal control approach for uncertainty input-delay systems based on adaptive dynamic programming

This paper concerned with a nearly optimal control approach based on adaptive dynamic programming technique to solve robust control problem of the neutral type time-delay systems, taking parameter uncertainties and input delay into account. Based on the neural network (NN)-based adaptive dynamic pro...

Celý popis

Uložené v:
Podrobná bibliografia
Vydané v:CACS : 2017 International Automatic Control Conference : 12-15 November 2017 s. 1 - 6
Hlavní autori: Lin, Yu-Chen, Chen, Hsin-Chang, Peng, Chih-Kai
Médium: Konferenčný príspevok..
Jazyk:English
Vydavateľské údaje: IEEE 01.11.2017
Predmet:
adaptive dynamic programming
Artificial neural networks
asymptotic stability
Cost function
Dynamic programming
Lyapunov-Razumikhin
Nearly optimal control
neural network
Optimal control
Robust control
Uncertainty
On-line prístup:Získať plný text
Tagy: Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
Popis
Shrnutí:This paper concerned with a nearly optimal control approach based on adaptive dynamic programming technique to solve robust control problem of the neutral type time-delay systems, taking parameter uncertainties and input delay into account. Based on the neural network (NN)-based adaptive dynamic programming and Lyapunov-Razumikhin theorems, the robust control design problem can be equivalently transformed into a nearly optimal control problem, and the amount of matched uncertainties are indirectly reflected in the performance index. A nearly optimal control is designed to approximate the costate function of the Hamilton-Jacobi-Isaaca (HJI) equation by NN-based adaptive dynamic programming scheme. By algebraic inequalities and appropriate uncertainty descriptions, sufficient conditions are derived under which not only the uncertain input-delay dynamical systems can achieve asymptotic stability, but also acquire the guaranteed level of performance for regulation. Simulation example is performed to demonstrate the effectiveness of the proposed approaches.
DOI:10.1109/CACS.2017.8284252