Metodologia de baixo custo para mapeamento geotécnico aplicado à pavimentação

Abstract

This paper presents a low-cost methodology for forecas%ng and mapping of CBR values (California Bearing Ra%o) of soils in the energies normal compression (CBR-N) and inter-mediate (CBR-I), which contribute to the decision-making process as to their use for paving purposes. GIS, Ar%ficial Neural Networks (ANN) and modeling techniques as well as biophysical and spa%al variables were used as explanatory of the modeled phenom-enon. The researched characteris%cs (pedology, geology, geomorphology, vegeta%on, hypsometry and posi%on) correlated with CBR values of soils in both energy compac%on. CBR data were extracted from pre-exis%ng projects and studies in the study area, in this case, the metropolitan area of Fortaleza (MAF). Thus, they were calibrated, validated and tested in many different ANN to find the two models best fit, for the genera%on of CBR-N es%mates and CBR-I, of the soil MAF from the studied biophysical variables. The geotechnical characteris%cs es%mated by these models enabled the development of two Neural Geotechnical Maps stra%fied to predict the values, CBR-N and CBR-I. The results show that ANN technique is promising to predict the mechanical proper%es of soils and can assist in making decisions regarding the use of these in road projects