2021 Fiscal Year Final Research Report
Extraction of high-order information by deep learning from environment monitoring data on greenhouses
| Project/Area Number |
19K06323
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Multi-year Fund |
|---|
| Section | 一般 |
|---|
| Review Section |
Basic Section 41040:Agricultural environmental engineering and agricultural information engineering-related
|
|---|
| Research Institution | Kindai University |
|---|
Principal Investigator |
|
|---|
| Project Period (FY) |
2019-04-01 – 2022-03-31
|
| Keywords | 培養液管理システム / エッジコンピューティング / 保守管理 / 環境制御システム / 高度異常監視 / Raspberry Pi / 深層学習モデル / UECS |
|---|
| Outline of Final Research Achievements |
For the advanced utilization of environmental measurement and control data, an intelligent monitoring system for inspection and maintenance of greenhouses using a deep learning (DL) model has been discussed. The DL model has learned normal greenhouse environment from the big-data record. A Raspberry Pi installed the learned DL model is attended to a greenhouse LAN with working the Ubiquitous Environment Control System (UECS), and compares predicted values by DL with measured real values every minute. The results of a demonstration tests confirmed that the system was able to detect delicate control abnormalities that are difficult to detect only by observing abnormal environmental measurement values and abnormal plant growth. A UECS node that automatically measures the amount of water and fertilizer absorbed by crops every minute was also designed and developed. The utilization as the training data sets for DL applications for advanced nutrient management in hydroponics is expected.
|
| Free Research Field |
植物生産工学、植物環境調節工学、農業情報工学
|
| Academic Significance and Societal Importance of the Research Achievements |
環境制御システムの普及で，日本で2018年に約1870憶個が記録されたと試算されるなど，園芸施設の環境計測制御電子記録データが大量に自動取得されている．しかし，過去データのグラフ化程度が主で，生産に資する活用がされていない．本研究では，深層学習の教師データとしてこれを活用し，環境制御不調を高度監視して，点検・保守に活用可能にした．開発システムは，高度な保守管理サービスとして農業クラウドビジネスでの事業化が期待できる．また，開発した深層学習施設環境モデルは，従来の重回帰モデルと比較して4～6倍高精度であり，時系列情報も不要である．施設園芸用環境モデルとして各種応用が今後期待できる．
|
