Utilização de algoritmo genético para a solução do problema de proporção de carvões para produção de coque metalúrgico

Abstract

In a highly competitiveness market environment companies are always trying reduce their costs and/or increase their profits. To reach these objectives a methodology used is the implementation of optimization tools integrated to the production process. Thus, it is expected to get maximum process efficiency with the minimum cost. An example of utilization this methodology is the solution of the problems of blending of coals to produce coke. The metallurgical coke is the most important raw material in a blast furnace, because it is the fuel that has highest cost and it is extremely important to stabilization process of the blast furnace.This study shows the development of a tool which is the propose of finding the best blending of metallurgical coals to produce coke. Normally the metallurgical coal optimizers are limited because they utilize only linear programming to solve this problem. However in the present study such formulation would not be efficient due to the existence of a restriction to the value of a feature of the coke, calculated using a neural network. Thus it is found a no linear problem which is solved through the use genetic algorithm.