Adaptive neural net feed forward system and method for adaptive control of mechanical systems
IPC분류정보
국가/구분
United States(US) Patent
등록
국제특허분류(IPC7판)
G05B-013/00
G05B-013/02
G05B-013/04
G05B-017/00
G06F-015/18
G06F-019/00
G06N-003/02
G06N-003/04
G06N-099/00
출원번호
US-0544014
(2009-08-19)
등록번호
US-8868221
(2014-10-21)
발명자
/ 주소
Mealy, James
출원인 / 주소
Marvell International Ltd.
인용정보
피인용 횟수 :
1인용 특허 :
13
초록▼
A feed forward system includes an input device configured to receive a desired response of the mechanical system and a trainable system configured to generate a feed forward signal that drives the mechanical system based on the desired response. The trainable system is trained based on a plurality o
A feed forward system includes an input device configured to receive a desired response of the mechanical system and a trainable system configured to generate a feed forward signal that drives the mechanical system based on the desired response. The trainable system is trained based on a plurality of measured responses of the mechanical system.
대표청구항▼
1. A feed forward control system for controlling a mechanical system, the feed forward control system comprising: an input device configured to receive a desired response of the mechanical system; anda trainable system configured for initial training based on a plurality of measured responses of the
1. A feed forward control system for controlling a mechanical system, the feed forward control system comprising: an input device configured to receive a desired response of the mechanical system; anda trainable system configured for initial training based on a plurality of measured responses of the mechanical system, wherein the trainable system, after being initially trained based on the plurality of measured responses, is configured to: receive the desired response from the input device; andin response to receipt of the desired response, generate a feed forward signal that drives the mechanical system, the feed forward signal being generated to produce an actual response of the mechanical system that matches the desired response received by the trainable system;a comparator configured to: receive the desired response from the input device and the actual response of the mechanical system; andgenerate an error signal based on a difference between the desired response and the actual response; anda feedback device configured to: receive the error signal from the comparator; andgenerate a feedback signal based on the error signal, wherein the feedback signal comprises a sum of a plurality of weighted adjustment values, wherein the trainable system is further configured to:receive the error signal from the comparator and the feedback signal from the feedback device; andadjust the feed forward signal based on the error signal and the feedback signal to generate an adjusted feed forward signal that drives the mechanical system. 2. The feed forward control system of claim 1, wherein the input device comprises a first delay device configured to delay transmission of the desired response to the trainable system, where the delay provided by the first delay device is programmable. 3. The feed forward control system of claim 2, wherein the feed forward signal is further transmitted to a second delay device, the second delay device configured to delay the feed forward signal and to transmit the delayed feed forward signal to the trainable system. 4. The feed forward control system of claim 1, further comprising: a summer device configured to generate an applied control signal, the applied control signal to be applied to the mechanical system based on the feed forward signal and the feedback signal,wherein a contribution of the feed forward signal to the applied control signal is greater than a contribution of the feedback signal. 5. The feed forward control system of claim 1, wherein the comparator is configured to determine the difference based on a comparison between the actual response of the mechanical system to the applied control signal and the desired response. 6. The feed forward control system of claim 1, wherein the trainable system is further trained with a plurality of controls signals applied to the mechanical system, each control signal of the plurality of control signals associated with a respective measured response of the plurality of measured responses. 7. The feed forward control system of claim 6, wherein the trainable system is further configured to generate the feed forward signal based on a control signal associated with one of the plurality of measured responses. 8. The feed forward control system of claim 6, wherein the plurality of control signals is indicative of a subset of a range of operation of the mechanical system. 9. The feed forward control system of claim 8, wherein the trainable system is further configured to identify a control signal not among the plurality of control signals, the control signal associated with a second desired response that fails to substantially match a measured response of the plurality of measured responses. 10. A feed forward control method for use with a mechanical system, the method comprising: receiving, with an input device, a desired response of the mechanical system;after a trainable system is trained based on a plurality of measured responses of the mechanical system: receiving, with the trainable system, the desired response from the input device; andin response to receiving the desired response, generating, with the trainable system, a feed forward signal that drives the mechanical system, the feed forward signal being generated to produce an actual response of the mechanical system that matches the desired response received by the trainable system;receiving, with a comparator, the desired response from the input device and the actual response of the mechanical system;generating, with the comparator, an error signal based on a difference between the desired response and the actual response;generating, with a feedback device, a feedback signal based on the error signal received from the comparator, wherein the feedback signal comprises a sum of a plurality of weighted adjustment values;receiving, with the trainable system, the error signal from the comparator and the feedback signal from the feedback device; andgenerating, with the trainable system, an adjusted feed forward signal that drives the mechanical system by adjusting the feed forward signal based on the error signal and the feedback signal. 11. The method of claim 10, further comprising delaying transmission of the desired response to the trainable system, where the delay is programmable. 12. The method of claim 11, further comprising delaying the feed forward signal and transmitting the delayed feed forward signal to the trainable system. 13. The method of claim 10, further comprising: generating, with a summer, an applied control signal applied to the mechanical system based on the feed forward signal and the feedback signal, wherein a contribution of the feed forward signal to the applied control signal is greater than a contribution of the feedback signal. 14. The method of claim 10, further comprising determining, with the comparator, the difference based on a comparison between the actual response of the mechanical system to the applied control signal and the desired response. 15. The method of claim 10, wherein the trainable system is further trained with a plurality of controls signals applied to the mechanical system, each control signal of the plurality of control signals associated with a respective measured response of the plurality of measured responses. 16. The method of claim 15, further comprising generating the feed forward signal based on a control signal associated with one of the plurality of measured responses. 17. The method of claim 15, wherein the plurality of control signals is indicative of a subset of a range of operation of the mechanical system. 18. The method of claim 17, further comprising identifying a control signal not among the plurality of control signals, the control signal associated with a second desired response that fails to substantially match a measured response of the plurality of measured responses.
연구과제 타임라인
LOADING...
LOADING...
LOADING...
LOADING...
LOADING...
이 특허에 인용된 특허 (13)
Kojima Yasuhiro (Hyogo JPX) Izui Yoshio (Hyogo JPX) Goda Tadahiro (Hyogo JPX) Kyomoto Sumie (Hyogo JPX), Control method using neural networks and a voltage/reactive-power controller for a power system using the control method.
Frerichs Donald K. (Shaker Heights OH) Kaya Azmi (Akron OH) Keyes ; IV Marion A. (Chagrin Falls OH), Method and procedure for neural control of dynamic processes.
Nomura Masahide (Hitachi JPX) Saito Tadayoshi (Hitachi JPX) Matsumoto Hiroshi (Ibaraka JPX) Shimoda Makoto (Katsuta JPX) Kondoh Masakazu (Hitachi JPX) Miyagaki Hisanori (Ohta JPX) Sugano Akira (Katsu, Process control method and system for performing control of a controlled system by use of a neural network.
Reid, Byron; Gutsmann, Volker; Zimmermann, Michael; Pienaar, Chris; Fages, Gaelle; Jardine, Peter, Value added pest control system with smart learning.
※ AI-Helper는 부적절한 답변을 할 수 있습니다.