数据驱动的多相交互冶金过程能耗优化方法研究及应用

项目名称： 数据驱动的多相交互冶金过程能耗优化方法研究及应用

项目编号： No.60874069

项目类型： 面上项目

立项/批准年度： 2009

项目学科： 金属学与金属工艺

项目作者： 阳春华

作者单位： 中南大学

项目金额： 36万元

中文摘要： 本项目深入分析了多相交互有色冶金过程工业运行数据的特点，研究了有色冶金过程异常数据检测、缺失数据补全、关键工艺参数软测量、测量数据在线校正等数据预处理方法，在此基础上，系统地开展面向节能降耗的数据驱动的有色冶金过程操作优化方法研究。首先定义了操作模式的形式化描述，构建了基于操作模式的有色冶金过程优化控制结构，提出了基于属性约简的操作模式提取方法以及类不平衡的优良操作模式库构建方法，建立了综合考虑生产指标和工况稳定的评价指标模型，研究了操作模式快速匹配搜索算法以获取优良操作模式，提出了基于优良操作模式与粒子群算法的操作参数智能优化方法，将优良操作模式映射为最优操作参数，进行生产过程中操作参数的最优调整，实现生产过程的优化运行。以铜闪速熔炼和锌湿法冶炼电解过程为应用对象，建立了基于能耗与炉况稳定性的铜闪速熔炼综合评价模型、基于电耗与电能费用的锌电解综合评价模型，提出了基于操作模式的铜闪速熔炼过程优化控制方法和基于数据的锌电解过程优化控制方法，工程应用验证了所提出方法的有效性。从而形成了较为系统的基于操作模式的有色冶金过程优化控制理论与方法，为基于数据的有色冶金过程优化控制提供理论基础。

中文关键词： 有色冶金过程；能耗优化；数据驱动；数据约简；模式匹配；

英文摘要： Based on the analysis of production data characteristics for nonferrous smelting process with multiphase interaction, the data preprocessing methods for nonferrous smelting process are investigated, including outlier detection, missing data completion, soft sensing of key process parameters, online data correction and so on, then the data-driven operational optimization method oriented to energy-saving and consumption-reducing for nonferrous smelting process is systematically studied. Firstly, the formal description of operational pattern is defined, and the operational-pattern-based optimization control structure for nonferrous smelting process is proposed. Then, the methods to extract operational pattern based on attribute reduction and the methods to create excellent operational pattern base using class-imbalanced data are presented. The evaluation model based on production index and stability is established, and a fast matching algorithm is proposed to search the excellent operation pattern. Finally, an intelligent optimization method based on excellent operational pattern and particle swarm algorithm is proposed to optimize the process parameters by mapping the excellent operational pattern to optimal operational parameters. The proposed methods are applied to copper flash smelting process and zinc electrolytic process. A synthetic evaluation model for copper flash smelting process based on energy consumption and stability of flash furnace is established, and the corresponding optimal control method based on operational pattern is proposed. A synthetic evaluation model for zinc electrolytic process based on power consumption and power cost is established, and the corresponding data-based optimal control method is proposed. The industrial application results demonstrate the effectiveness of the proposed methods. Therefore, the operational-pattern-based optimal control theory of nonferrous smelting process is founded, which provides theoretical principle of data-based optimization for nonferrous smelting process.

英文关键词： nonferrous smelting process; energy consumption optimization; data-driven; data reduction; operational pattern matching