Rock Identification by Functional Networks

摘要

A civil engineer has to deal mostly with soils, rocks, timber, steel and concrete. In a great majority of civil engineering projects, plans and designs involve the soils and rocks almost directly. As engineering properties of rocks are rather generalized terms, study and determination of all such properties of roches that are relevant to engineering application of these materials either after excavation from natural beds or without extraction as in-situ conditions is very essential. The identification of rocks based on physical and micro properties using standard regression analysis is difficult. In this paper, functional network (FN) as well as multiple linear regression techniques are used to predict the rock type using the physical and micro properties. Unlike neural networks (NN), functions replace weights in functional networks. The prediction accuracy by both these methods is quite good.