摘要
针对当前选矿生产过程中磨矿参数难以实时测量,导致的无法对磨矿流程控制进行有效实时优化的问题,提出了一种基于改进狮群算法和BP神经网络的磨矿浓度预测方法。传统的狮群算法(Lion Swarm Optimization, LSO)存在容易陷入局部最优解和局部搜索能力弱的问题,通过改变狮王更新方式的同时加入衰减因子,来提升全局搜索能力,避免陷入局部最优解,对于母狮和学习狮的更新方式引入了动态学习策略,来优化局部搜索能力,并在仿真测试中验证了改进算法的优越性。同时利用改进的算法代替BP神经网络中的梯度下降法,来搜索最优的权值和阈值,提升BP神经网络收敛速度和收敛精度。试验结果表明,改进狮群算法优化的BP神经网络(Improved Lion Swarm Optimization-BP,ILSO-BP)在此预测问题上有更好的效果。
In order to solve the problem that the grinding parameters are difficult to be measured in real time in the current beneficiation production process,and the grinding process control can not be effectively optimized in real time,agrinding concentration prediction method based on improved Lion Swarm Algorithm and BP neural network was proposed.The traditional Lion Swarm Optimization(LSO)was easy to fall into local optimal solution and the local search ability is weak,by changing the update method of the lion king and adding the attenuation factor,to improve the global search ability and avoid falling into local optimal solution,for the update method of the lioness and learning lion,the dynamic learning strategy was introduced to optimize the local search capability.And the superiority of the improved algorithm was verified in the simulation test.At the same time,the improved algorithm was used to replace the gradient descent method of BP neural network to search the optimal weights and thresholds,and improve the convergence speed and accuracy of BP neural network.The experimental results showed that the Improved Lion Swarm Optimization-BP(ILSO-BP)neural network had a better effect on this prediction problem.
作者
葛子源
王庆凯
邹国斌
杨佳伟
刘道喜
GE Ziyuan;WANG Qingkai;ZOU Guobin;YANG Jiawei;LIU Daoxi(Beijing General Research Institute of Mining and Metallurgy,Beijing 100160,China;BGRIMM Technology Group,Beijing 100160,China;State Key Laboratory of Automatic Control Technology in Mining and Metallurgy Process,Beijing 102628,China)
出处
《有色金属（选矿部分）》
CAS
北大核心
2023年第5期59-66,共8页
Nonferrous Metals（Mineral Processing Section）
基金
国家重点研发计划项目(2020YFE0201100)
甘肃省科技计划项目(20ZD7WC010)。
关键词
狮群算法
衰减因子
动态学习
BP神经网络
磨矿浓度预测
Lion Swarm Algorithm
reducation factor
dynamic learning
BP neural network
grinding concentration prediction
