Um Modelo de previsão do módulo de resiliência dos solos no estado do Ceará para fins de pavimentação

Abstract

The development of models to support the transport infrastructure design has established itself as an alternative to obtain data and information on the geotechnical characteristics of the soil in a given region. One of the techniques that have been successful in generating the estimates is the Artificial Neural Networks (ANN).The neural modeling has allowed both the prediction of different geotechnical soil attributes as their locations with relative accuracy for a given study area. It is known that the methods of empirical-mechanistic pavement design depend on their application, prior knowledge of some of these characteristics. Obtaining geotechnical soil information, such as AASHTO classification, CBR and resilient modulus (RM), imply high financial cost. Thus, this thesis proposes the use of artificial intelligence techniques to generate models to estimate the RM of soil in order to be used in methods of pavement design, for which this information is essential. It is proposed to perform conventional geotechnical testing (Particle size analysis of soils and Compaction) to obtain data that will allow the proposed model, aiming to estimate the resilient behavior of soils in the state of Ceará. From the results, estimates were generated that can be integrated into the pavement design methods for the study area, reducing the financial costs of the projects and the execution time. The results showed that the ANN are able to predict with good accuracy, with 0.984 correlation coefficient, RM values of the soil, showing that the use of neural models to predict the RM to the empirical-mechanistic design of pavements is possible. This technique allows the use of the models generated for the design of pavements, where there is lack of information or lack of funding for road design.