基于生成对抗网络的浮选加药过程建模  

Modeling for Flotation Reagent Addition Processes Based on Generative Adversarial Network

作　　者：李希 胡文静 刘浪[1,2] 易嘉闻 赵林 LI Xi;HU Wenjing;LIU Lang;YI Jiawen;ZHAO Lin(School of Information Science and Engineering,Hunan Institute of Science and Technology,Yueyang 414006,China;Research Center of Machine Vision and Artificial Intelligence,Hunan Institute of Science and Technology,Yueyang 414006,China)

机构地区：[1]湖南理工学院信息科学与工程学院,湖南岳阳414006 [2]湖南理工学院机器视觉及人工智能研究中心,湖南岳阳414006

出　　处：《湖南理工学院学报（自然科学版）》2021年第3期1-8,共8页Journal of Hunan Institute of Science and Technology(Natural Sciences)

基　　金：湖南省自然科学基金项目(2019JJ40110,2019JJ40104);湖南省教育厅科学研究项目(18B349,19A201,20A214);湖南省研究生科研创新项目(CX20190932,CX20190930)。

摘　　要：针对矿物浮选过程中浮选药剂添加量与泡沫图像特征之间关系复杂且难以准确建模的问题,提出一种基于生成对抗网络的浮选加药过程智能建模方法.利用生成对抗网络的样本对抗思想,通过判别器计算加药过程中泡沫图像变化前后的特征差异,生成器学习到不同药剂添加量情况下的泡沫图像特征.利用工业现场采集的浮选加药数据进行实验验证,结果表明,所提浮选加药过程建模方法能够有效构建加药量和泡沫图像之间的关系,在不同工况下模型得到的加药后泡沫图像特征均与实际的加药后泡沫图像一致.Considering that it is difficult to model the relationship between flotation reagent dosage and froth image features,an intelligent modeling method for flotation reagent addition processes based on generative adversarial network(GAN)was proposed.This method used the sample confrontation idea of the generative adversarial network,the discriminator calculated the feature difference between the real froth image and the generated froth image,and the generator learned the froth image feature under different reagent dosage conditions.The flotation reagent addition data collected on the industrial scene was used for experimental verification and model performance analysis.The results show that the flotation dosing process modeling method proposed in this paper can effectively construct the relationship between reagent dosage and froth image,and the froth image features obtained by the model after reagent addition are consistent with the actual froth image features under different working conditions.

关 键 词：生成对抗网络 加药 泡沫浮选 泡沫图像 矿物浮选过程 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] TD923[自动化与计算机技术—控制科学与工程]