Hybrid dynamic scheduling framework using layered fuzzy inference and RBF neural network

Abstract

Effective dynamic scheduling is an essential element in the process of intelligent road construction. The primary goal of this paper is to outline a two stage framework of dynamic scheduling for construction using layered fuzzy inference and radial basis function (RBF) neural network. The layered fuzzy inference presents an initial model which embeds the experts' knowledge by Zadah fuzzy theory and decision fusion. The RBF neural network adaptively adjusts the parameters of the initial model during the operation process. The experiment of the actual engineering problem shows that the scheduling results accord with the human knowledge and the training of the model needs less time compared with BP neural network. The proposed hybrid framework has been integrated in the practical asphalt road construction scheduling system.