Abstract
In this contribution, difficulties of closed-loop adaptive training of a neural controller (MC) are discussed, and a "half closed-loop" training strategy is proposed when the sign of the derivative of the plant‘s output to its control input is known. Two cost functions for one step ahead control are chosen as our training cost function alternatively to overcoats the difficulties encountered when either of them is used only. We show that in certain cases, the two cost functions will determine the same controller. At first the proposed strategy is implemented with a multilayer feedforward neural network (MFNN) when the order of the plant is known. Then plants with unknown order are also considered, and an inner recurrent neural network (IRNN) is employed to implement the proposed strategy. Simulations in this paper have shown the validity of the new idea.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.