DYNAMIC LOAD ESTIMATION IN SILAGE HARVESTING

Abstract

There are many sources of disturbances for on-the-go weighing. This study was conducted to develop a new methodto remove those disturbances and to estimate more accurate dynamic masses in silage harvesting. A mathematically simpleprocedure was developed using vertical movement information from a low-cost accelerometer, in addition to conventionalinstrumentation using load cells for mass data, and was tested using a small scale model weighing bin and a commercial silagewagon. Clear similarities were found between the patterns of low-pass filtered mass and acceleration data when both wereobtained simultaneously from the same harvesting system. Multiplication factors for acceleration data were calculated sothat the differences between the mass data and the multiplied acceleration data could be minimized. Thus, subtracting themultiplied acceleration from the mass data corresponded to deleting load cell disturbances due to vertical movements. Asmall-scale model weighing system was used to apply the developed approach to known dynamic masses. Mass estimationshowed less than 20 g measurement errors for experimental loads. The same method was applied to a commercial silage wagonwith harvested silage mass data in the 0 to 7,000 kg range. The proposed method showed that remaining error magnitudeswere reduced by 39% to 56% and standard deviations were reduced by 53% to 68 % with respect to the results of low-passfiltering and moving average.