Loading...

Control a Double Inverted Pendulum Using Active Learning Method

Ghomi Rostami, Mohammad | 2011

642 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 42027 (05)
  4. University: Sharif University of Technology
  5. Department: Electrical Engineering
  6. Advisor(s): Bagheri Shouraki, Saeid
  7. Abstract:
  8. There are lots of researches performed to investigate human being learning and try to model it. But, in case of huge complexity of its structure and function, almost none of so-far learning algorithms can model human learning process and, indeed, performance of human brain in real problems is much more than these algorithms. Moreover, in many cases, human is not aware or informed about internal structure or mathematical model of the present system he is working on; a child learning to walk has no information about his neuromuscular system and his brain slowly learns how to convey his little body by choosing the proper timing of muscle contractions. One of the accomplished researches in order to make artificial brain and exploit human brain abilities is Active Learning Method (ALM). Active Learning Method is an adaptive-recursive fuzzy learning algorithm which is inspired by some behavioral features of human brain and active learning ability of human. In this algorithm, a Multi-Input Multi-Output (MIMO) system is interpreted as a fuzzy combination of some simpler Single-Input Single-Output (SISO) subsystems. The knowledge of each subsystem consist of its input/output characteristic and its degree of importance, is extracted by a fuzzy operator named as Ink-Drop Spread (IDS) and finally is combined by other subsystems in inference layer of the algorithm. This operator is processing engine of ALM algorithm and regulates the the fuzzy-like resolution of practical data in input/output planes. In procedure of controller design using ALM, while combining SISO subsystems, we try to estimate the next state of system and then, by asserting the input to change the state as required, move the system to its desired set-point. In this thesis, design procedure of ALM controller is proposed in an extended manner to higher order systems. First, we simulate an inverted pendulum (single link) to test the ALM controller and then, a double inverted pendulum. At the end, we tried to control the double inverted pendulum without any knowledge of upper link. The results show the simple linguistic rules of expertness are interpreted by ALM and the final controller can control the complex and non-linear systems in acceptable range.
  9. Keywords:
  10. Expertise Extraction ; Active Learning ; Fuzzy Controller ; Double Inverted Pendulum

 Digital Object List

  • محتواي پايان نامه
  •   view

 Bookmark

No TOC