1994 Fiscal Year Final Research Report Summary
Fuzzy Logic Control of Toroidal Plasmas
| Project/Area Number |
04680020
|
|---|
| Research Category |
Grant-in-Aid for General Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Research Field |
プラズマ理工学
|
|---|
| Research Institution | National Institute for Fusion Science |
|---|
Principal Investigator |
YAMAZAKI Kozo National Institute for Fusion Science, Professor, 大型ヘリカル研究部, 教授 (50115648)
|
|---|
| Project Period (FY) |
1992 – 1994
|
| Keywords | Fusion / Toroidal Plasma / Real-Time Control / Fuzzy Control / Feedback Control |
|---|
| Research Abstract |
This research aims at applying fuzzy logic to the plasma control, at clarifying the applicability of this algorism, and at searching for the prospect of the future fusion reactor control.
The control objets in magnetically confined toroidal plasmas are plasma current, plasma position, plasma cross-section, plasma density, heating power and so on. In addition to those, in the reactor, neutron production rate and fusion power output should be controlled. Until now classical PID (Proportional, Integrated and Derivative) control schemes are adopted in the fusion research. Since the machine system becomes bigger, a new flexible and reliable control concept is being required.
The author are continuing the plasma control research in both tokamak and helical systems, and developing computer codes for plasma modeling. As for fuzzy control research, a mini-max center-of-gravity fuzzification method with triangular membership function is applied to the plasma current control in helical plasma.
During this fiscal year (Heisei 6th) a fuzzy control algorism was added to the LHD plasma control R&D machine. This control scheme will be applied to the fusion experimental machine and fusion power reactor, and will contribute to the systematization of fusion control. At the beginning of this research, only fuzzy control logic was surveyed, but wide and basic investigations on various intelligent controls have recently been started, including neural network combined with fuzzy logic and new concepts (Chaos Engineering and Generic Algorism).
|
