Scale-up support technology for newly developed products using existing production data
| Project/Area Number |
15K06554
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Multi-year Fund |
|---|
| Section | 一般 |
|---|
| Research Field |
Reaction engineering/Process system
|
|---|
| Research Institution | Kyoto University |
|---|
Principal Investigator |
Kano Manabu 京都大学, 情報学研究科, 教授 (30263114)
|
|---|
| Project Period (FY) |
2015-04-01 – 2018-03-31
|
| Project Status |
Completed (Fiscal Year 2017)
|
| Budget Amount *help |
¥4,940,000 (Direct Cost: ¥3,800,000、Indirect Cost: ¥1,140,000)
Fiscal Year 2017: ¥1,300,000 (Direct Cost: ¥1,000,000、Indirect Cost: ¥300,000)
Fiscal Year 2016: ¥1,690,000 (Direct Cost: ¥1,300,000、Indirect Cost: ¥390,000)
Fiscal Year 2015: ¥1,950,000 (Direct Cost: ¥1,500,000、Indirect Cost: ¥450,000)
|
|---|
| Keywords | スケールアップ / ベイズ的最適化 / 転移学習 / 製剤 / ガウス過程回帰 / 操業条件最適化 / ソフトセンサー / 仮想計測 / 変数重要度 / 因果推論 |
|---|
| Outline of Final Research Achievements |
We aim to develop a new method that can optimize operating conditions of commercial-scale equipment to achieve scale-up from pilot-scale equipment even when only a small number of experiments can be performed with commercial-scale equipment. The proposed method, combined task Bayesian optimization (CTBO), uses not only data of a target task, e.g., a commercial-scale plant, but also data of a source task, e.g., a pilot-scale plant. CTBO determines new operating conditions in the target task sequentially by BO while information of the source task is exploited by transfer learning. CTBO was compared with BO and LW-PLS + jDE (locally weighted partial least squares + self-adaptive differential evolution) through their applications to a pharmaceutical granulation process. CTBO remarkably outperformed the other methods. CTBO is expected to be useful not only for scale-up but also for technology transfer.
|
Report
(4 results)
Research Products
(14 results)
