Soft computing for blast furnace gas system pressure based on an improved fuzzy model

摘要

The stability of blast furnace gas (BFG) system is of great importance in steel manufacturing process. This paper proposes a multi-objective hierarchical genetic method for building a fuzzy system to measure the pressure of BFG network in complex industrial environments. In order to improve the accuracy of the model, the fuzzy system is divided into four layers with the optimization target of mean absolute percentage error and root mean square error, including the input layer, the membership layer, the rule base layer and the fuzzy system layer. Then, a coding strategy for each layer is designed and an objective function for calculating the fitness value of each individual is established to achieve the purpose of co-evolution for each layer. Moreover, a Levenberg-Marquart Bayesian regularization algorithm is employed to solve the overfitting problem in the modeling process. The experimental results using a series of practical production data collected from a steel plant show the validity of the proposed method, and the established T-S fuzzy model could provide scientific support for the energy management in steel production.