Um método para localização e estimação das características geotécnicas dos solos da região metropolitana de Fortaleza-CE para fins de pavimentação

Abstract

This research focuses on the development of a method, aimed to predicting and positioning the geotechnical characteristics of soils that may contribute to the process of decision making of its use for paving purposes. Were used Geoprocessing and Artificial Neural Networks (ANN) modeling techniques, as well as spatial and biophysical variables of the phenomena modeled. The characteristics studied (pedology, geology, geomorphology, vegetation, altimetry and position) were correlated with the estimated geotechnical variables (TRB Classification and CBR) for soils from the metropolitan region of Fortaleza, Ceará (RMF). Three models of ANNs were developed calibrated, validated and tested. Two of these models were dedicated to generating estimates of CBR in the normal (CBR-N) and intermediate (CBR-I) compaction modes. The third model was developed to generate estimates of the geotechnical characteristics of the soils from the RMF Classification TRB. The geotechnical characteristics estimated by these models enabled the preparation of Neural Geotechnical Maps, stratified for values of CBR-N, CBR-I and TRB Classification. The maps produced and all the survey information was made available on a Web Geographic Information System (Webmapping), thus allowing its use in road projects and future academic studies, both to download the maps and to generate estimates for RMF. In addition, provided to the Webmapping a geotechnical receiver points, to allow recalibration of future models in an attempt to improve the quality of the estimates that currently is more than 90% accuracy rate.