An Approach to Deconvoluting Grain-Flow within a Conventional Combine Harvester using a Parametric Transfer Function

Grain flow measured at the flow sensor in a commercial yield monitoring system is considered to be the convoluted result of grain entering the harvester at one time being mixed by the internal threshing and transport processes with grain entering at other times. A transfer function describing the flow of grain within a conventional combine harvester was used to deconvolve the recorded signal using Fast Fourier Transformation (FFT). This process of time series analysis was shown to reposition the yield data in a more accurate manner than the simple time-delay process currently employed. The deconvolution increases the variability in the yield data from 10% to 17% which more closely resembles the variation in crop yield expected from small area samples. The spatial structure of yield variation obtained from small area samples was also shown to be more closely estimated by the deconvoluted data set, however the process does display sensitivity to noise introduced by mechanical/electrical sources and procedural errors.