Research of coke ratio prediction based on concept lattice and genetic algorithm

The concept lattice is adopted as a tool of attribute reduction to reduce the redundant factors affecting coke ratio in this paper. On this basis, in order to solve the blindness and random problems in the parameters of artificial selection in support vector machine (SVM), this paper adopts genetic algorithm to optimize the penalty parameter C, kernel function parameters γ and insensitive loss coefficient e of support vector machine and put forward the prediction model based on the concept lattice and genetic algorithm optimization for support vector machine (Con-GA-SVM), which is applied to forecast coke rate of blast furnace. Through comparative experiment, this algorithm has better performance than PSO-SVM prediction model and Grid-SVM model.