Ameliorated moth-flame algorithm and its application for modeling of silicon content in liquid iron of blast furnace based fast learning network

Moth-Flame Optimization (MFO) algorithm is a widely used nature-inspired optimization algorithm. However, for some complex optimization problems, such as high dimensional and multimodal problems, the MFO may fall into the local optimal solution. Hence, in this paper an ameliorated Moth-Flame Optimization (AMFO) algorithm is presented to improve the solution quality and global optimization capability. The key features of the proposed algorithm are the Gaussian mutation produce flames and the modified position updating mechanism of moths, which can improve the ability of MFO to jump out of local optimum solutions. In addition, opposition-based learning is adopted to initialize the population. The AMFO algorithm is compared with 9 state-of-the-art algorithms (such as Levy Moth-Flame Optimization (LMFO), Grey Wolf Optimization (GWO), Sine Cosine Algorithm (SCA), Heterogeneous Comprehensive Learning Particle Swarm Optimization (HCLPSO)) on 23 classical benchmark functions. The comparative results show that the AMFO is effective and has good performance in terms of jumping out of local optimum, balancing exploitation ability and exploration ability. Furthermore, the AMFO is adopted to optimize the parameters of fast learning network (FLN) to build the prediction model of silicon content in liquid iron for blast furnace, and simulation experiment results from field data show that the root mean square error of the AMFO-FLN model is 0.0542, hit ratio is 91 and the relative error is relatively stable, the main fluctuation is between-0.1 and 0.1; compared with other ten silicon content in liquid iron models, the AMFO-FLN model has better predictive performance. (C) 2020 The Author(s). Published by Elsevier B.V.