OPTIMAL NEURAL NETWORK CONTROL
1993
Abstract This paper proposes an unsupervised learning control system based on multi-layered feed-forward neural networks. A new strategy is developed for training neural network controller such that an overall performance index over a period of time is optimized. Therefore, the system tends to learn an optimal control law without the requirement of a priori knowledge of plant dynamics. The new control strategy is evaluated by using the cart-pole system. Simulation results show its strong self-learning ability and robustness. Comparisons with linear controller and another commonly used neural network method also demonstrates the superiority of our system.
Keywords:
- Correction
- Cite
- Save
- Machine Reading By IdeaReader
3
References
1
Citations
NaN
KQI