Fuzzy - Neural Network and Cost Optimization using Genetic Algorithm in Modeling Marshal Stability of Stabilized Soils with Emulsion Bitumen and Cement

Intelligent methods for predicting some quantities are powerful tools for optimization of predicting models these days. The bitumen emulsion is able to easily mix with wet soil so that the bitumen may be easily distributed and scattered in the soil. The aggregates with PI<6 are quite suitable for stabilization of soils with bitumen emulsion. In this research, two aims are followed, one is Marshal Stability modeling for samples stabilized with emulsion bitumen and cement for roads sub-layers, and the other is cost optimization of using these binding materials in roads construction. 170 stabilized samples were made and tested for training the fuzzy neural network. The Marshal Stability of these samples were modeled by fuzzy neural network and also with using genetic algorithm, the roads execution costs were optimized. For modeling, the MATLAB software programming was used. The results showed that, with presented model, the construction costs decrease considerably.The overall conclusions of this research are summarized as follows:1. The fuzzy neural network presented, is able to determine the Marshall Stability of the samples stabilized by bitumen emulsion and cement with an accepted error %, and establishes a logical relation between the portions of used cement and emulsion.2.The fuzzy neural network model is a suitable tool for modeling and estimating the natural problems. 3.The genetic algorithm optimization model presented in this research is able to optimize the performance costs of soil stabilization by cement and bitumen emulsion, and consequently reduces total costs of road projects, considerably.4.The presented diagram for Marshall Stability – cement and emulsion content is able to specify the required cement and emulsion portions as compared to Marshall Stability.