赤铁矿选矿过程运行指标的软件模型设计  

Software Model Design of Operating Index of Hematite Ore Dressing Process

作　　者：张帆[1] ZHANG Fan(Network Teaching and Research Department of Computer and Software College,Shanxi Institute of Technology,Xi’an 710300)

机构地区：[1]陕西国防工业职业技术学院计算机与软件学院网络教研室

出　　处：《微型电脑应用》2019年第8期63-65,共3页Microcomputer Applications

基　　金：陕西国防工业职业技术学院课题项目(Gfy 16-19)

摘　　要：针对赤铁矿选矿过程的特点和技术指标的精度要求,通过计算机技术结合布谷搜索(CS)算法和磁滞切换(HS)策略的混合式多传感器磨削粒度的软测量建模方法构建了选矿过程指标软件。首先,根据磨削过程的工艺特点和大量实验数据,推导出磨削粒度机理软测量模型;同时,建立了BP神经网络软测量模型和小波神经网络(WNN)软测量模型。建立了基于磁滞切换策略的混合型多传感器软件模型,根据每个采样时刻的切换性能指标,软件自动选择最优模型作为当前的预测模型;最后软件采用搜索算法完成对迟滞切换策略的性能参数进行优化。仿真结果表明,该软件模型有较好的泛化结果和预测精度,能够满足磨矿分级过程的实时控制要求。According to the characteristics of the ore dressing process of hematite and the precision requirements of the technical indicators,software of the ore dressing process indicators is constructed through the combination of computer technology with the soft measurement modeling method of the grinding granularity of hybrid multi-sensor combined,the cloth valley search(CS)algorithm,and hysteresis switching(HS)strategy.Firstly,the soft measurement model of grinding particle size mechanism is derived based on the process characteristics and experimental data.Meanwhile,a BP neural network model and a wavelet neural network(WNN)model are established for soft measurement.The hybrid multi-sensor software model based on hysteresis switching strategy is realized.According to the switching performance index at each sampling time,the software automatically selects the optimal model as the current prediction model.Finally,the software uses the search algorithm to optimize the performance parameters of the delayed switching strategy.The simulation results show that the software model has good generalization results and prediction accuracy,and can meet the real-time control requirements of grinding classification process.

关 键 词：选矿过程 软测量模型 BP神经网络 遥感探测 检测点评估模型 

分 类 号：TP392[自动化与计算机技术—计算机应用技术]