Bart Missotten

A Review of Combine Sensors for Precision Farming

To maximize economic return from agricultural production units, costs have to be minimized and benefits maximized. For grain, kernel yield and quality have to be maximized while the use of seeds, fertilizer, herbicides and fungicides have to be optimized.The best location to evaluate productivity levels, by measuring yield and quality of grain and straw, is the combine harvester. Moreover, other grain quality characteristics like density or test weight can be determined for use as an evaluation tool. In this paper, an overview is given of the past and current research toward the evaluation of currently available commercial sensors (e.g., for measuring grain yield and grain moisture content) as well as toward the development of new sensors (e.g., grain protein content and straw yield).

Performance Evaluation of a Three-dimensional Optical Volume Flow Meter

Optical sensors in combine grain elevator housings have been used for volume flow measurements. Several parameters affecting the accuracy of optical volume flow detection are described. Parameters considered were the slope, the kernel type, the moisture content, and the clean grain elevator capacity. A positive linear relationship was found between the optical sensor output and the volume flow. Using the output of only one horizontal sensor to explain the volume flow of all the runs showed maximum deviations from the sample regression that ranged up to 13%. Using four optical sensors, two mounted transverse to the elevator chain and two mounted perpendicular to the chain and symmetrically on both sides of the chain, the maximum deviation from the sample regression was reduced to 9%. These results were obtained after reduction of the gap between the paddles and the housing.

ON-LINE TRACKING OF SEPARATION PROCESSES IN COMBINE HARVESTERS

Separation and cleaning units of harvesters are typically exposed to changing crop and field conditions and are characterized by a highly variable, static process behavior. A general algorithm is proposed to track the static, non-linear relation between two time shifted signals. This algorithm is based on a stochastic Newton optimization algorithm and can be implemented on standard software platforms that are currently available on commercial combine harvesters. A stochastic Newton algorithms has been constructed for a single parameter, exponential feedrate-grain loss model. This single parameter tracking algorithm has been applied on a measured data set. The sitespecific interpretation of the estimated exponential parameter shows the significant relation with the local field slope. Therefore, on-line mapping techniques can be useful to tune the combine settings on the basis of an on-the-go estimated separation efficiency.