| Project/Area Number |
24500184
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Multi-year Fund |
|---|
| Section | 一般 |
|---|
| Research Field |
Intelligent informatics
|
|---|
| Research Institution | Chubu University |
|---|
Principal Investigator |
|
|---|
| Project Period (FY) |
2012-04-01 – 2015-03-31
|
| Project Status |
Completed (Fiscal Year 2014)
|
| Budget Amount *help |
¥5,200,000 (Direct Cost: ¥4,000,000、Indirect Cost: ¥1,200,000)
Fiscal Year 2014: ¥1,170,000 (Direct Cost: ¥900,000、Indirect Cost: ¥270,000)
Fiscal Year 2013: ¥2,730,000 (Direct Cost: ¥2,100,000、Indirect Cost: ¥630,000)
Fiscal Year 2012: ¥1,300,000 (Direct Cost: ¥1,000,000、Indirect Cost: ¥300,000)
|
|---|
| Keywords | 制限付き学習法 / 昆虫脳 / 組み込み用学習アルゴリズム / MPPTマイクロコンバータ / 太陽電池 / learning on a budget / 組み込み用学習エンジン / モデルベース制御 / 昆虫脳のアルゴリズム / マイクロコンバータ / 追加学習 / kernel perceptron / 一般回帰ニューラルネットワーク / incremental learning |
|---|
| Outline of Final Research Achievements |
We proposed a tiny learning machine : "limited general regression neural network (LGRNN)", which is inspired by insect's brain. LGRNN learns new instances one by one, and selects important instances, and store them into a limited capacity. This ability is realized by choosing one of several learning options adaptively by estimating an evaluation function, which predicts expected error due to the execution of each learning option.
The proposed method was applied to a micro-converter of photovoltaics, which improves the electricity generation ability of solar panels. The system consists of a conventional hill climbing mechanisum for finding the maximum power point(MPP), and LGRNN. Normally, the device tracks the MPP using the conventional method, while the LGRNN learns the relationships between the current magnitude of solar-irradiation, tempeleture and MPP. When the solar-irradiation changes abruptly, however, LGRNN yields the predicted MPP for controling the chopper circuit quickly.
|
