Quantification of tacit knowledge of machine operation and discrimination of operational conditions by skilled farmers

2022 Fiscal Year Final Research Report

• Project/Area Number: 20K06323
• 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: University of the Ryukyus
• Principal Investigator: Mitsuoka Muneshi 琉球大学, 農学部, 准教授 (60437770)
• Project Period (FY): 2020-04-01 – 2023-03-31
• Keywords: 農業機械の作業情 / 複数時系列データ / 特徴量抽出 / 機械学習 / 機械の異常検出 / 作業状態の分類

Outline of Final Research Achievements

In this study, the objective is to extract features that can be used to　understand the working conditions of agricultural machinery only from vibration time-series data during operation, and to quantify these conditions. To this end, I attempted to extract features that contribute to understanding　differences in work conditions based on vibration time-series data and to　classify work conditions by machine learning.
As a result, time-series data obtained from inertial sensors (3-axis translational acceleration and 3-axis rotational angular velocity) installed in a rigid part of the machine were the most useful in understanding the operating conditions of the machine. In addition, machine learning using the measured vibration time-series data for 16 patterns of different work and field conditions for tractor rotary tillage operations showed that the machine was able to discriminate with an accuracy of about 95 %.

Free Research Field

農業情報工学

Academic Significance and Societal Importance of the Research Achievements

一般的に熟練作業者ほど圃場の土壌や作物の状態，機械から伝わる振動・騒音および作業精度等の情報から作業状態を総合的に判断して操作を行っている。また，作業中に機体を通じて作業者に暴露される力覚データは機械の状態を作業者に知覚させる重要な情報であり、機械の不具合や異常検知における判断・意思決定に大きく影響を及ぼす。
そこで本研究では、計測された多変数データを用いた異常の早期検出手法を開発することで作業適期中の機械の故障による経済的損失の低減、メンテナンス性や使用耐久年数および機械の安全性の向上に極めて高い寄与が期待できる。