基于PSO和CBR优化粒度的磨矿过程设定控制  

Setting Control for Size Optimization in Grinding Process Based on PSO and CBR

作　　者：刘晓青[1] 李芳[1] 程全[1] 李晋[1] 杨静[2] 

机构地区：[1]周口师范学院机械与电气工程学院,河南周口466001 [2]周口师范学院物理与电信工程学院,河南周口466001

出　　处：《控制工程》2017年第3期594-599,共6页Control Engineering of China

基　　金：国家自然科学基金(11547227);河南省自然科学基金(152300410134);河南省科技攻关计划项目(132102210179.132102210577;142102210599);高校微课教学活动在网络学习空间中的实效研究(15A880023);河南省教育技术装备和实践教育研究项目(GZS310)

摘　　要：磨矿产品粒度直接关系到选矿厂的金属回收率、精矿品位等技术指标,针对磨矿过程滞后时间长、参数时变严重、强非线性、强耦合等特性,采用案例推理技术(CBR)实现磨矿粒度优化。相似度计算是CBR中案例检索的关键环节,直接关系到案例检索的精度。传统采用欧式距离计算相似度的方法,通常假设案例各属性的权重固定且相互独立,而该假设往往不能满足实际应用。针对此问题,提出一种基于粒子群优化算法(PSO)的自学习相似度计算方法,并将其引入案例推理中,构成粒度指标智能优化设定系统,并联合常规的基础控制系统,构建了磨矿过程优化设定控制系统,保证磨矿过程整体优化稳定运行。应用到某大型选矿厂的磨矿流程,取得了明显成效,具有推广应用价值。The particle size of the grinding process is directly related to metal recovery, concentrate grade and other technical indicators of the concentrator. In allusion to the grinding process with characteristics such as long lag time, time-varying parameters, strong nonlinearity, strong coupling and so on, the case-based reasoning （CBR） method is used to realize the optimization of the grind size. Similarity computation is a key link in the process of case retrieval in CBR, which directly relates to the accuracy of the case retrieval. Traditional similarity is obtained by using the Euclidean distance method, which usually assumes fixed weights of each attribute and these attributes are independent of each other. But the hypothesis often cannot meet the practical applications. Aiming at this problem, this paper presents a self-learning method for similarity calculation, based on particle swarm optimization （PSO） algorithm. Applied to the case-based reasoning, the intelligent optimal setting system for the granularity index is constituted, To ensure the overall optimization and stability in the grinding process, the optimal control system is structured, combining the intelligent optimal setting system for granularity index with the regular basic control system. Applied to the grinding process of a large-scale concentrator, the proposed method achieves remarkable results, and is worth promoting.

关 键 词：磨矿粒度 案例推理(CBR) 粒子群算法(PSO) 优化控制 

分 类 号：TP27[自动化与计算机技术—检测技术与自动化装置]