基于数据模型的铅冶炼氧化炉原料配比优化

Optimization of raw material ratio for lead smelting oxidation furnace based on data model

底吹连续处理铅基固废工艺具有多变量、非线性、强耦合、大滞后等特点,基于机理方法进行建模与优化存在困难。对此本文提出了基于数据驱动的熔炼炉原料配料模型,实现关键运行参数的优化控制。首先,基于化验与过程历史数据,使用神经网络建立原料成分与熔炼炉关键工艺指标间的关系模型;在此基础上,,应用粒子群搜索算法,由熔炼炉理想工况指标搜索确定原料中各成分的最优配比;最后,将配料问题建模为含非线性约束的多目标优化问题,并使用SLSQP求解。集成上述建模优化算法,开发了相应的熔炼炉原料管理系统。

The bottom-blowing continuous treatment for lead-based solid waste has the characteristics of multivariability,nonlinearity,strong coupling and large lag,which cause difficulties for mechanism based modelling and optimization.To solve these problems,this paper proposes a data-driven raw material blending model for the smelting furnace,which achieves optimized control for key operating parameters.Firstly,based on laboratory and process historical data,the relationship between raw material composition and key process indicators of the smelting furnace is established by applying neural network;on this basis,the Particle Swarm Optimization algorithm is applied to solve the optimal ratio of each component in the raw material from the ideal operating conditions;finally,the ingredient problem is formulated as a multi-objective optimization problem with nonlinear constraints and then solved by SLSQP.Integrating the above modeling and optimization algorithms,a corresponding raw material management system has been developed.