Title :
Approximating optimal state feedback using neural networks
Author :
Mcdermott, Wesley ; Athans, Michael
Author_Institution :
Dept. of Electr. Eng. & Comput. Sci., MIT, Cambridge, MA, USA
Abstract :
The training and usage of multilayer neural networks on discontinuous (e.g. bang-bang) feedback control problems are discussed. Training sets are created from optimal open loop trajectory information and a heuristic for trimming the base data set is presented. A priori knowledge about solution trajectories is seen to improve the training process
Keywords :
feedforward neural nets; neurocontrollers; optimal control; state feedback; base data set; discontinuous feedback control; multilayer neural networks; optimal open loop trajectory; optimal state feedback; Computer science; Cost function; Feedback control; Multi-layer neural network; Neural networks; Open loop systems; Optimal control; Reflection; State feedback; State-space methods;
Conference_Titel :
Decision and Control, 1994., Proceedings of the 33rd IEEE Conference on
Conference_Location :
Lake Buena Vista, FL
Print_ISBN :
0-7803-1968-0
DOI :
10.1109/CDC.1994.411510
