Adaptive critic based neural networks have been found to be powerful tools in solving various optimal control problems. The adaptive critic approach consists of two neural networks which output the control values and the Lagrangian multipliers associated with optimal control. These networks are trained successively and when the outputs of the two networks are mutually consistent and satisfy the differential constraints, the controller network output produces optimal control. In this paper, we analyze the mechanics of convergence of the network solutions. We establish the necessary conditions for the network solutions to converge and show that the converged solution is optimal.
S. N. Balakrishnan and X. Liu, "Convergence Analysis of Adaptive Critic Based Optimal Control," Proceedings of the 2000 American Control Conference, 2000, Institute of Electrical and Electronics Engineers (IEEE), Jan 2000.
The definitive version is available at https://doi.org/10.1109/ACC.2000.879538
2000 American Control Conference, 2000
Mechanical and Aerospace Engineering
Keywords and Phrases
Lagrangian Multipliers; Adaptive Control; Adaptive Critic Method; Convergence; Dynamic Programming; Learning; Learning (Artificial Intelligence); Necessary Conditions; Neural Networks; Neurocontrol; Neurocontrollers; Optimal Control
Article - Conference proceedings
© 2000 Institute of Electrical and Electronics Engineers (IEEE), All rights reserved.