From sensor values to a map: accuracy of spatial modelling methods in agricultural machinery works

摘要

Spatial modelling of the farm machinery operations has several steps that can cause errors and inaccuracies for the modelled maps. Maps are used for precision farming, and the inaccuracies will have a direct effect on the quality of precision farming.The aim of this study was to determine and to evaluate the extent of the error factors that have effects on spatial modelling. Most common field farming machines including harvesters, seeding machines, harrows, sprayers and spreaders were examined and analysed from the spatial modelling perspective. Interacting factors were determined and classified. The effects of the factors were simulated and determined using simulated or previously driven and collected field data. Results were accomplished by incorporating or leaving out different factors, and also focusing on present common spatial modelling practises. In-depth studies focused on a combine harvester and a fertilizer spreader. A combine harvester was chosen because of its popularity in precision farming (PF). The fertilizer spreader was chosen due to most complicated spatial modelling requirements. This study demonstrates the possible significance of different factors that have an effect on the map generation. At this stage, the model consists offive main components: Measurement values, lags in measurement system, dynamic tractor-implement combination, changing work pattern and interpolation. After the raw data is collected, the unconnected lags in the measurement system could change the longitudinal location for several metres. All the measurement systems will have some uncertainty with the log timing. Dynamic tractor-implement combination generates more error sources. Uneven field surface and curved driving lines urges the need of accurate modelling. Changes in the temporal spatial distribution of the work are rarely taken account of but generate errors again with curved driving lines and with variable rate application (VRA) actions. Finally the interpolation can smooth the results or can also lose important information.