BCOISOA-BP网络在磨矿粒度软测量中的应用
发布时间：2018-10-17 15:00
【摘要】：传统人群搜索(SOA)算法通过计算搜索方向、搜索步长和搜寻更新个体位置三个步骤进行寻优.它的缺点在于计算量大,种群之间信息交流少,导致寻优速度慢.针对人群搜索算法存在的缺点,本文提出二项交叉算子改进人群搜索算法(BCOISOA)对其改进.在计算搜索步长方面,本文采用随机数与最大函数值位置乘积判断子群位置,进而提高全局寻优计算速率.在更新位置方面,本文提出二项交叉算子加强种群之间的联系,避免在更新搜索方向过程中,算法因局部最优而导致过早收敛,进而达到快速、准确寻找最优解的目的.本文将以上二项交叉算子改进人群搜索-BP神经网络算法应用在二段式磨矿过程中,实现磨矿粒度在线软测量.仿真结果表明,与人群搜索算法和粒子群算法进行比较,二项交叉算子改进人群搜索算法收敛速度更快,预测精度最高,满足对磨矿粒度实时检测的要求.
[Abstract]:Traditional crowd search (SOA) algorithm is optimized by calculating search direction, searching step size and searching update individual position. Its disadvantages are that it has a large amount of computation and less information exchange between populations, which results in a slow speed of optimization. In view of the shortcomings of the crowd search algorithm, this paper proposes a binomial crossover operator to improve the population search algorithm (BCOISOA). In the aspect of calculating search step size, the product of random number and maximum function value is used to determine the position of subgroup, and the calculation rate of global optimization is improved. In the aspect of updating position, this paper proposes a binomial crossover operator to strengthen the relationship between populations, so as to avoid premature convergence of the algorithm due to local optimum in the process of updating search direction, and then achieve the purpose of finding the optimal solution quickly and accurately. In this paper, the improved crowd search (BP) neural network algorithm of the above two crossover operators is applied to the two-stage grinding process to realize on-line soft measurement of grinding particle size. The simulation results show that compared with the crowd search algorithm and particle swarm optimization algorithm, the binomial crossover operator can improve the convergence speed of crowd search algorithm faster and the prediction accuracy is the highest, and meet the requirements of real-time detection of grinding granularity.
【作者单位】： 河北工业大学控制科学与工程学院;北京科技大学自动化学院;
【基金】：河北省高等学校科学技术研究资助项目(ZD2016071)
【分类号】：TD921.4;TP18
本文编号：2277053
[Abstract]:Traditional crowd search (SOA) algorithm is optimized by calculating search direction, searching step size and searching update individual position. Its disadvantages are that it has a large amount of computation and less information exchange between populations, which results in a slow speed of optimization. In view of the shortcomings of the crowd search algorithm, this paper proposes a binomial crossover operator to improve the population search algorithm (BCOISOA). In the aspect of calculating search step size, the product of random number and maximum function value is used to determine the position of subgroup, and the calculation rate of global optimization is improved. In the aspect of updating position, this paper proposes a binomial crossover operator to strengthen the relationship between populations, so as to avoid premature convergence of the algorithm due to local optimum in the process of updating search direction, and then achieve the purpose of finding the optimal solution quickly and accurately. In this paper, the improved crowd search (BP) neural network algorithm of the above two crossover operators is applied to the two-stage grinding process to realize on-line soft measurement of grinding particle size. The simulation results show that compared with the crowd search algorithm and particle swarm optimization algorithm, the binomial crossover operator can improve the convergence speed of crowd search algorithm faster and the prediction accuracy is the highest, and meet the requirements of real-time detection of grinding granularity.
【作者单位】： 河北工业大学控制科学与工程学院;北京科技大学自动化学院;
【基金】：河北省高等学校科学技术研究资助项目(ZD2016071)
【分类号】：TD921.4;TP18
【相似文献】
相关期刊论文 前8条
1 A.B.博尔特尼科夫;E.B.奥尔洛娃;B.M.巴利亚斯诺娃;万起;;关于确定合理的磨矿粒度问题[J];国外金属矿山;1991年12期
2 丘德镳;怎样确定最佳磨矿粒度[J];矿冶工程;1991年03期
3 王新华;桂卫华;王雅琳;阳春华;;混合核函数支持向量机的磨矿粒度预测模型[J];计算机工程与应用;2010年12期
4 李茂林;崔瑞;王非;向文娟;曾凡霞;;阶段磨选磨矿粒度划分的理论分析与计算[J];矿冶工程;2011年01期
5 何桂春,毛益平,倪文;基于多神经网络的磨矿粒度软测量模型[J];金属矿山;2005年02期
6 郭素红;刘威;;某钾盐矿磨矿粒度对浮选效果影响的试验研究[J];有色金属(选矿部分);2009年04期
7 张剑锋;;梅山原矿及二次分级溢流降硫试验研究[J];现代矿业;2009年10期
8 杨晓峰;陈景明;;回收赤铁矿尾矿工艺中合适的磨矿粒度及设备研究[J];矿业工程;2013年06期
相关会议论文 前1条
1 杨晓峰;陈景明;;浮选尾矿再选工艺中的合适的磨矿粒度与设备研究[A];晋琼粤川鲁冀辽七省金属（冶金）学会第二十一届矿业学术交流会论文集[C];2014年
相关硕士学位论文 前1条
1 刘德明;基于支持向量回归的磨矿粒度软测量及其软件实现[D];辽宁科技大学;2012年
，本文编号：2277053
本文链接：https://www.wllwen.com/kejilunwen/kuangye/2277053.html
