Sensing Miscanthus Stem Bending Force for Maximizing Throughput Rate in a Disk Mower-Conditioner

Abstract

One of the reasons for relatively high biomass harvesting cost is challenges in adjusting the ground speed of harvesting machines with respect to the yield level within a field. A real-time biomass yield sensor that can predict the yield in front of a machine could be a useful tool to control ground speed. It was hypothesized that the force required to bend Miscanthus stems is a reliable predictor of biomass yield. Based on this novel concept, a stem bending force-sensing system was developed and field tested with a disk mower-conditioner. A bale-specific method, segmenting the field area from which a bale was formed, was developed to correlate sensed bending force and Miscanthus yield. The measured bending force showed a logarithmic relationship (R 2 = 0.80) with Miscanthus yield. The average error in predicting balespecific yield was 10.3% for training data and 12.9% for validation data. The average error in predicting plot yield was 3.4% for training plots and 10.0% for validation plots. Using the developed logarithmic correlation model, yield maps were also generated. For the specific case analyzed, a proper control strategy to maximize throughput rate (mass per unit time) would be to either operate the mower-conditioner at the maximum feasible ground speed (9 km h-1) or at the maximum achievable throughput rate (60 Mg h-1). The yield-sensor controlled machine would result in 44.2% higher field capacity, 41.3% higher throughput rate, and 31.2% lower mowing-conditioning cost for the specific case analyzed compared to the operator-controlled machine. Studies are needed to extend the stem bending force-sensing concept to other thickstemmed crop harvesting machines, such as sugarcane harvesters and coppice harvesters.