学术文献库

Yield monitors on harvesters are a key component of precision agriculture. Mass flow estimation is the critical factor to measure, and having this allows for field productivity analysis, adjustments to machine efficiency, and cost minimization by ensuring trucks are filled maximally without exceeding weight limits. Several common technologies used on grain harvesters, including impact plate sensors, are accurate enough on combines to be valuable but suffer from issues such as drift. Sugarcane is composed of a mixture of billets and trash, which is a very dispersed material with much less consistency than grains. In this study, a 3d point cloud approach is used to estimate volume, from which a calibration factor is derived [density] to translate to mass. The system was proved in concept in a controlled environment using bamboo, achieving an R2 of 97.4% when fitting average volume flow per test against average mass flow after correcting for bulk density changes with volume. The system was also tested on field data, which was collected from nearly 1700 wagon loads from the southern U.S. and Brazil over the course of 3 seasons in both green and burnt cane. Results indicated that the concept is very robust with good accuracy, having seasonal CVs for density values ranging from 6.9% to 16.2%. The camera concept proves relatively robust to environmental conditions. The same approach could be used in sugar beets, potatoes or other sparse/non-flowing crops with highly varying material properties, where traditional mass flow sensors do not work.