Data-driven Operational-pattern Optimization for Copper Flash Smelting Process
基于数据驱动的铜闪速熔炼过程操作模式优化及应用

Considering the difficulties of modeling,online-measurement of technical indexes, and optimal control incopper flash smelting process, a data-driven operational-patternoptimization method is presented. Firstly, the copper flash smeltingprocess is analyzed, basic concepts about data-drivenoperational-pattern are defined and the frame of data-drivenoperational pattern optimization is proposed. Secondly, thedata-driven prediction models of matte temperature, matter grade andratio of Fe to SiO2 are established, the overall evaluation modelof flash smelter is proposed and operational-pattern optimizationfor copper flash smelting process is described. Thirdly, anoptimized operational-pattern base is constructed based on lots ofindustrial running data and the overall evaluation model. Then, amatching algorithm combining fuzzy C-means cluster with chaos pseudoparallel genetic algorithm is proposed to mine an optimaloperational pattern from the optimized operational-pattern base toimplement the optimal control of the smelting process. The practicalrunning results in copper flash smelting process show itseffectiveness.