Multi-Layer Encoding Genetic Algorithm-Based Granular Fuzzy Inference for Blast Furnace Gas Scheduling

Abstract

Abstract: A timely and effective scheduling of byproduct gas system in steel industry is very significant for cost reduction and environment protection. In this study, a granular-based fuzzy inference model is proposed for blast furnace gas (BFG) scheduling, in which the fluctuation characteristics of the flow of the byproduct gas users is considered for unequal-length data partition, and the time warping normalization (TWN) is exploited to equalize the data segments into the granules with same length. By applying fuzzy clustering, the adjustment amount of each gas user and the system adjustment amount are granularized to the form of fuzzy sets. Furthermore, a fuzzy inference model is built up to describe the relationships between the two variables and thus establish the gas scheduling rules. To improve the precision of fuzzy inference, a multi-layer coded genetic algorithm is proposed to determine the prior’s parameters including the major influential users and their corresponding clustering numbers. A case study applied in a steel plant in China demonstrates that the proposed scheduling model can guarantee a higher accuracy and make the operation of blast furnace gas system safe and stable.