Sunil Mathanker and Alan Hansen 2016-05-03 01:49:35
Yield sensors integrated with RTK-GPS (real-time kinematic global positioning system) have been beneficial in site-specific crop management. However, variations in crop throughput raise concerns about the correlation between predicted yield and actual yield at specific locations. Furthermore, time delays make many available yield sensors unsuitable for real-time control of harvester parameters, such as travel speed. A harvester controlled with a yield sensor has potential for improved throughput and reduced harvesting cost, and efforts have been made toward integrating yield sensors with harvester control for grain crops. However, a literature survey found few studies that investigated the use of real-time sensors for predicting biomass yield. The stem bending concept We were instrumental in developing two real time biomass yield sensors. Our first sensor measured biomass volume to predict biomass yield while harvesting Miscanthus. Our second sensor uses a different approach. It measures stem bending force to predict biomass yield. The stem bending concept was initially developed for Miscanthus, a potential bioenergy crop. In a more recent study, we extended the concept to Napiergrass, another potential bioenergy crop. Crop stem bending has been incorporated into many harvester designs, such as sugarcane, coppice, and corn harvesters. Miscanthus stems are bent to facilitate their cutting and feeding into the machine. Typically, the stems are bent by a push bar or knockdown roller that is placed ahead of the stem cutting device. The stem bending force at the push bar or roller can be used to sense the biomass yield. To achieve this, we sandwiched load cells between two parallel pipes that acted as a modified push bar for Miscanthus or as a modified knockdown roller for Napiergrass, performing a similar function as the original push bar or roller. The bending force sensed at the push bar was normalized with the area harvested within a one-second duration. The normalized bending force was then correlated with the actual yield at that specific location. Strong correlations were found for Miscanthus (R2 = 0.73) and Napiergrass (R2 = 0.92). Since then, the stem bending sensor has been used for yield mapping, and it can also be used for controlling a harvester. Yield mapping A yield map for Miscanthus, generated with the aid of a stem bending sensor, is shown in the illustration below. The yield levels are divided into three categories: low (20 Mg ha-1). The bale locations are also shown. The yield map shows that the bottom rows produced two or three bales per row, and the top rows produced one bale per row. From the number of bales per row, it can be inferred that the top rows contained a low-yielding crop while the bottom rows contained a highyielding crop. The actual yield levels were used to validate the yields predicted from the stem bending force. The average yield prediction accuracy was within 10% for Miscanthus and within 11.8% for Napiergrass. Controlling a harvester In addition to mapping yield, the stem bending sensor shows potential for controlling the operating parameters of a harvester. A possible control approach is based on adjusting the harvester’s ground speed using the yield levels determined from the stem bending sensor. This approach is constrained by two operating parameters: the harvester’s maximum feasible ground speed and maximum achievable throughput rate. If the yield levels are lower than the typical threshold, then the harvester will be operated in speed control mode. If the yield levels are higher than the threshold, then the harvester will be operated in throughput control mode. After harvesting a Miscanthus field while controlling the harvester with the stem bending sensor, our analysis showed a 41.3% increase in throughput and a 31.2% decrease in harvesting cost. This approach of sensing the stem bending force at the point of initial machine contact offers some major advantages in yield mapping, and it creates opportunities for optimizing machine performance and productivity. The stem bending sensor can also be adapted to other thick-stemmed crops. However, some limitations of this approach still need to be addressed. For example, adjusting the harvester’s ground speed can change the sensed bending force because the bending force is a function of ground speed. Crop varietals and growth variations present another limitation. The ability of the harvester to respond to sudden fluctuations in yield also needs to be investigated. Most likely, operator input would be required to respond to changes in biomass properties caused by moisture, dew, and other factors. Overall, mapping biomass yield, and controlling the biomass harvester, based on the crop’s stem bending force is a very promising approach, and further studies should help us overcome these limitations for field implementation. ASABE member Sunil Mathanker, Assistant Professor, Department of Agricultural and Biosystems Engineering, University of Puerto Rico at Mayaguez, Mayaguez, Puerto Rico, USA, firstname.lastname@example.org. ASABE Fellow Alan Hansen, Professor, Department of Agricultural and Biological Engineering, University of Illinois, Urbana, USA, email@example.com. The authors wish to acknowledge that the work reported in this article was funded by the Energy Biosciences Institute (EBI) within a program titled “Engineering Solutions for Biomass Feedstock Production.”
Published by ASABE. View All Articles.