|TOPIC||Approximate Dynamic Programming Based Control|
|SPEAKER||Professor S. N. Balakrishnan
Department of Mechanical and Aerospace Engineering
Missouri University of Science and Technology
|DATE||31 Jan 2011 (Monday)|
|TIME||02:30 pm to 03:30 pm|
|VENUE||Level 2, Social Robotics Lab, I3 Building,
Interactive Digital Media Institute, 21 Heng Mui Keng Terrace, NUS.
|Approximate dynamic programming formulation implemented with an Adaptive Critic (AC) based neural network (NN) structure has evolved as a powerful alternative technique that eliminates the need for excessive computations and storage requirements needed for solving the Hamilton-Jacobi-Bellman (HJB) equations. A typical AC structure consists of two interacting NNs. In this paper, a novel architecture, called the Single Network Adaptive Critic (SNAC) is used to solve control-constrained optimal control problems. Only one network is used that captures the mapping between states and the cost function. This approach is applicable to a wide class of nonlinear systems where the optimal control (stationary) equation can be explicitly expressed in terms of the state and costate variables. A non-quadratic cost function is used that incorporates the control constraints. Necessary equations for optimal control are derived and an algorithm to solve the constrained-control problem with SNAC is developed. Convergence of the network training is discussed. Benchmark nonlinear systems are used to illustrate the working of the proposed technique. Extensions to optimal control-constrained problems in the presence of uncertainties are also considered. Two aerospace tracking applications —-an aircraft and a space vehicle with model and parametric uncertainties will be shown.
The last part of the presentation deals with implementation of the intelligent controllers in a distributed parameter system. Distributed parameter systems are driven by partial differential equations and there are very few implementable solutions in the literature. We show that the adaptive critic based neurocontroller can successfully be implemented in a heat diffusion system to result in desired temperature profiles.
|Professor S. N. Balakrishnan received the Ph.D. degree in aerospace engineering from the University of Texas at Austin, Austin.
He has been with the University of Missouri—Rolla, Rolla, since 1985. Currently, he is a professor with the Department of Mechanical and Aerospace Engineering. His nonteaching experience includes work as a lead engineer in the Space Shuttle program, Fellow, Center for Space Research at the University of Texas at Austin, summer Faculty Fellow at Air Force Research Laboratory, Eglin, FL, and engineer, Indian Space Program. His research interests include areas of system theory and applications. His current research uses neural networks and classical methods in the identification and robust control of missiles, airplanes, rockets, and other “interesting” systems. His research has been sponsored by the National Science Foundation (NSF), the Air Force, the Naval Surface Warfare Center, the Army Space and Missile Defense Command, and NASA.
Dr. Balakrishnan is a member of Sigma Gamma Tau. He is an Associate Fellow of the American Institute of Aeronautics and Astronautics (AIAA)