콤바인 수확량 모니터링 시스템의 현장 실증 및 정확도 분석

초록

Accurate yield monitoring is essential in precision agriculture, as it offers detailed insights into crop productivity and facilitates data-driven decision-making. This study evaluated the field performance of an integrated yield-monitoring system installed on a DH6135-A (6-row) combine harvester. The system includes a load-cell-based grain-tank weight sensor, an impact-plate grain-flow sensor, and a GNSS-based 1-meter grid yield-mapping algorithm. Field experiments were conducted across seven rice paddies in Dangjin, Republic of Korea, during the 2024 harvest season. Sensor outputs were calibrated using linear regression models, and real-time measurements were refined through filtering, time-lag correction, and weight-scaling procedures. The sensor-based yield estimates were then compared with actual field-measured yields to assess system accuracy. The results indicated that the error rates between measured and monitored total yields ranged from -4.3% to +1.7%, with a mean absolute percentage error (MAPE) of approximately 1.5%. A strong linear relationship was established between monitored and actual yields, evidenced by a correlation coefficient of 0.99 and an R² value of 0.98. These findings demonstrate that the proposed yield-monitoring system performs reliably under Korean rice-growing conditions and is effective for generating spatial yield maps that enhance precision nutrient management, productivity analysis, and site-specific decision-making in rice production systems.

키워드