농업기계공학 · 한국농업기계학회

Research Article

Development and validation of a dynamic simulation model for estimating power requirements of a self-propelled garlic collector
자주식 마늘 수집기의 소요동력 추정을 위한 동역학 시뮬레이션 모델 개발 및 검증

Jin-Woo Park, Young-Woo Do, Yi-Seo Min, Seok-Pyo Moon, Young-Jo Nam, Seung-Gwi Kwon, Wan-Soo Kim
박진우, 도영우, 민이서, 문석표, 남영조, 권승귀, 김완수

The aim of this study is to develop a multi-body dynamics simulation model for predicting the power requirements of a self-propelled garlic … + READ MORE

The aim of this study is to develop a multi-body dynamics simulation model for predicting the power requirements of a self-propelled garlic collector and to validate the model through comparison with field-measured load data under two soil conditions. A simulation model of a 40.2 kW garlic collector, comprising a driving part, a collection part, and a transport part, was developed using RecurDyn. Field experiments were conducted at Hapcheon (silt loam) and Uiseong (clay loam), and the measured rotational speeds were applied as simulation input conditions. Under the Hapcheon condition, relative errors between simulated and measured power requirements were 6.1% for the driving part, 11.4% for the collection part, and 6.0% for the transport part, with a total power error of approximately 1.8%. Under the Uiseong condition, the errors for the main components and the total required power were all within 5%. The prediction errors in the driving part were mainly associated with crawler-ground interaction under different soil strength conditions, whereas the errors in the collection and transport parts were attributed to simplified contact, friction, chain tension, hydraulic load characteristics, and field operating conditions in the mechanical models. These results indicate that the developed model can predict the total power requirement of the garlic collector with acceptable accuracy and can serve as an engineering tool for engine capacity evaluation and power transmission optimization. - COLLAPSE

PDF

References Information

30 June 2026
Research Article

Hanwoo face-based individual identification using deep learning
딥러닝을 이용한 한우 안면 기반 개체 인식

Tae-eui An, Soo-Hyun Cho, Dae-Hyun Lee
안태의, 조수현, 이대현

Individual identification in livestock farms is a fundamental technology for precision livestock management, history tracking, and productivity management. Conventional identification methods, such … + READ MORE

Individual identification in livestock farms is a fundamental technology for precision livestock management, history tracking, and productivity management. Conventional identification methods, such as ear tags, sensors, and RFID devices, require physical attachment to animals and have limitations related to device loss, damage, and management burden. To overcome these limitations, camera-based non-contact identification technologies have received increasing attention. However, in practical livestock environments, existing cattle are repeatedly shipped out and new cattle are continuously introduced, requiring a technology that can reflect newly added individuals in the identification system. Therefore, in this study, we proposed a deep learning-based individual registration and identification framework using Hanwoo facial images to register new cattle with a small number of facial images. The proposed framework learns facial features from existing cattle, generates representative features from registration images of new cattle, and identifies individuals by comparing the similarity between input image features and registered features. In addition, the effects of the number and viewpoint composition of registration images on new-cattle identification performance were analyzed. Results showed a test accuracy of 98.68% for the initial identification model, and the highest overall average accuracy and new-cattle accuracy were 94.8% and 91.0%, respectively. Although identification performance varied depending on registration conditions, the proposed framework confirmed that newly introduced cattle can be reflected in the identification system using only a small number of registration images. Based on the results, the proposed approach is expected to improve the field applicability and operational efficiency of non-contact Hanwoo identification systems. - COLLAPSE

PDF

References Information

30 June 2026