A Bayesian-Influence Model for Error Probability Analysis of Combine Operations in Harvesting

Abstract

Harvesting is one of the most important agricultural operations because it captures the value from the entire cropping season. In modern agriculture, grain harvesting has been mechanized through the combine harvester. A combine harvester enables highly productive crop harvesting. Combine harvesting performance depends on the highly variable skill of combine operators and associated operator error. An approach was developed to analyze the risk of the combine harvesting operation as it relates to operator error. Specifically, a risk analysis model was built based on a task analysis from operator interviews and estimates of the probability of operator error. This paper employs a Bayesian approach to assess risks in combine operation. This approach applies a Bayesian Belief Network to agriculture operations, which represents a new application for this risk analysis tool. Sensitivity analysis of different errors and operator skill levels was also performed. The preliminary results indicate that a reduction of human operator action errors can substantially improve the outcomes of the human-machine interaction.