基于深度学习算法的湿法冶金资源回收效率提升方法设计及研究  

Design and Research of Hydrometallurgical Resource Recovery Efficiency Improvement Method Based on Deep Learning Algorithm

作　　者：宋玉安[1] 赵伟 SONG Yu’an;ZHAO Wei(School of Materials and Engineering,Jiyuan Vocational and Technical College,Jiyuan 459000,China)

机构地区：[1]济源职业技术学院材料工程学院,河南济源459000

出　　处：《湿法冶金》2025年第1期125-131,共7页Hydrometallurgy of China

摘　　要：为进一步提高湿法冶金资源回收率,解决资源回收流程控制的智能化、自动化控制程度不高的问题,提出了一种采用Transformer模型进行金属浸出率预测,再采用Distributional Q-function改进DQN模型进行湿法冶金金浸出率最大化的湿法冶金流程控制方法。结果表明:该系统控制方法能有效提升湿法冶金过程中金属浸出率的预测准确率;基于Distributional Q-function改进DQN模型能有效降低资源回收率最大化模型的迭代计算时间。该法能有效提高某工厂湿法冶金资源回收率。In order to further improve the recovery rate of hydrometallurgical resources and solve the problem that the intelligent and automatic control degree of resource recovery process control is not high,a hydrometallurgical process control method is proposed,which uses Transformer model to predict metal leaching rate and then uses Distributional Q-function to improve DQN model to maximize gold leaching rate.The results show that the system control method can effectively improve the prediction accuracy of metal leaching rate in hydrometallurgy process.Improving the DQN model based on Distributional Q-function can effectively reduce the iterative calculation time of the model with maximum resource recovery rate.The method can effectively improve the recovery rate of hydrometallurgical resources in a certain plant.

关 键 词：Transformer模型 最优化 Distributional Q-function DQN模型 资源回收 

分 类 号：TF803.21[冶金工程—有色金属冶金]