APPLICATION OF ARTIFICIAL NEURAL NETWORKS TO SKIDDER TRACTION PERFORMANCE

摘要

An artificial neural network simulating the tractive performance of a rubber-tired skidder operating on soft organic soil in the Coastal Plain region of North Carolina, U.S.A. was modeled for three tire sizes inflated at each of three inflation pressures (69, 103, and 172 kPa). The neural network (4-5-3-1) demonstrated good generalization of the pull-load relationship when presented with data not used in network training. The output of the network was in close agreement with the equations fitted using least squares methods for the actual pull-load data where the line pull increased linearly with the applied tree load.