著者

植松 英穂 竹田 辰興 西尾 成子

出版者

一般社団法人 日本物理学会

雑誌

日本物理学会誌 (ISSN:00290181)

巻号頁・発行日

vol.56, no.6, pp.395-402, 2001-06-05 (Released:2008-04-14)

参考文献数

6

---

日本において,制御核融合の研究が開始されて約50年経った.当時,天体,原子核,素粒子,宇宙線,放電,溶接などの分野の研究者たちによって核融合を志向する研究が始まった.そのとき,まず研究体制が議論され,さしあたって基礎研究を進めることで合意が得られた.その後,実験装置の大型化が進められるようになり,特に,この十数年で国際協力としての研究開発が盛んになった.本稿では,研究開発の巨費化がはじまる前の時代に焦点を当て,日本の制御核融合研究の跡をたどる.

著者

廣岡 伸治 服部 克巳 西橋 政秀 紺 晋平 竹田 辰興

出版者

The Institute of Electrical Engineers of Japan

雑誌

電気学会論文誌. A, 基礎・材料・共通部門誌 = The transactions of the Institute of Electrical Engineers of Japan. A, A publication of Fundamentals and Materials Society (ISSN:03854205)

巻号頁・発行日

vol.131, no.9, pp.691-697, 2011-09-01

---

The ionospheric anomaly prior to the 2007 SouthernSumatra earthquake (M8.5) was observed by GPS receivers around Sumatra islands. In this paper, to investigate the three-dimensional structure of electron density in Ionosphere, a tomographic approach (Residual Minimization Training Neural Network; RMTNN) has been performed. The results of the tomographic approach are consistent with those of total electron content (TEC) approaches. We found that the significant decreases take place in the heights of 250-400 km, especially at 330 km height. But the height which gives the maximum electron density is not changed. The obtained structure is that the decreased region exists in the southwest side of Integrated Electron Content (IEC) (400-550 km altitudes) and in the northern side of IEC (250-400 km altitudes). Global tendency of the decreases area is expanded to the east with an altitude and it is concentrated in the southern hemisphere of over the epicenter. These resultsshow that the high capability of RMTNN method for the estimation of the ionospheric electron density distribution possibly associated with earthquake.

著者

竹田 辰興 彌政 敦洋

出版者

社団法人 プラズマ・核融合学会

雑誌

プラズマ・核融合学会誌 (ISSN:09187928)

巻号頁・発行日

vol.78, no.9, pp.842-856, 2002 (Released:2005-12-08)

参考文献数

57

---

Applications of neural networks to data analysis and control of fusion plasmas are reviewed. First, a brief introduction to the general features of a neural network is presented, where the neural network is considered as a continuous mapping device, a classification device, a statistical processing device, and a time series predicition device. Then, the applications of neural networks to the research field are explained where the problems to be solved are classified a sfitting function, shaping an experimentally obtained spectrum, analyzing equilibrium quantity, prediction, tomography, and control problems. Throughout the article, we restrict ourselves to description of applications of multi-layer neural networks.