|Sudduth, Kenneth - Ken|
Submitted to: ASAE Annual International Meeting
Publication Type: Other
Publication Acceptance Date: 7/13/2000
Publication Date: N/A
Interpretive Summary: Precision agriculture is a crop management strategy which seeks to address within-field variability. An important aspect of precision agriculture is yield mapping. Most commercial yield mapping systems for grain crops use a sensor where the impact of grain against a plate is measured. For some other crops, such as potatoes or peanuts, a weighing device is used. With weighing-type systems, errors due to machine vibrations and travel across rough ground can cause problems. In this research, we adapted a weighing system mounted in a grain combine for on-the-go weighing. We used advanced signal processing techniques called Fourier analysis to remove errors in the weight readings caused by vibrations. We were able to remove over 80% of the error present in the initial weight signal. In addition to yield measurement, these techniques may find other applications in precision agriculture, such as measuring and mapping the amount of fertilizer applied. This research may benefit researchers who are working on development of weighing systems and equipment manufacturers who are interested in incorporating such systems into their products.
Technical Abstract: A weigh bin, mounted on load cells in the grain tank of a commercial combine, was evaluated for stationary and mobile grain weight measurement. A reference load cell with a known mass provided compensation for on-the-go weighing. Frequency analysis was used to determine the dominant frequencies in the bin and reference signals. Non-coherent frequencies were removed. Accelerations were calculated from the reference mass and used to compensate the bin weights. These data were then compared to data from a commercial yield monitor. The weigh bin provided accurate, repeatable measurements when used in a static mode. The raw data from on- the-go weighing was highly variable with standard deviations between 30 and 40 kg. Over half of this standard deviation was caused by disturbance frequencies greater than 3 Hz. Very little grain flow information was contained in the higher frequencies. Coherence measurement between reference mass data and weigh bin data showed a strong correlation only fo frequencies below 3 Hz. Application of a running median to lowpass filtered, compensated data resulted in a standard deviation between 1 to 7 kg. The data from this process compared well with data from a commercial yield monitor but still needs some refinement before being considered equivalent to the commercial system. Removing more of the noise from the signal through filtering or an improved compensation algorithm would result in a highly sensitive on-the-go weighing system.