NEURAL IDENTIFICATION OF EARTHMOVING MACHINERY'S PRODUCTIVITY

摘要

This paper deals with application of artificial neural networks to the identification of the productivity of earthmoving machinery consisting of excavators and haulers. Studies of cycle times for selected sets of machines were carried out. A feed-forward multilayer error back propagation network with a conjugate gradient algorithm was constructed using data, such as the technical parameters of the machinery, the category of the road for hauling the excavated material and the machinery's productivity, obtained from a queuing system model simulating the work of sets of machines. The neural network was used for productivity prediction. A methodology for the neural identification of productivity and the results showing the selected neural network as suitable for this purpose are presented.