欢迎访问过程工程学报, 今天是

过程工程学报 ›› 2023, Vol. 23 ›› Issue (9): 1300-1312.DOI: 10.12034/j.issn.1009-606X.222390CSTR: 32067.14.jproeng.222390

• 研究论文 • 上一篇    下一篇

基于神经网络预测的锌挥发率影响机制分析

昝智1,2， 张晨牧2*， 伍继君1*， 石垚2， 刘朗明3， 刘卫平3， 庄才备3   

  1. 1. 昆明理工大学冶金与能源工程学院，云南 昆明 650093 2. 中国科学院绿色过程与工程重点实验室，战略金属资源绿色循环利用国家工程研究中心，中国科学院过程工程研究所，北京 100190 3. 株洲冶炼集团股份有限公司，湖南 株洲 412007
  • 收稿日期:2022-10-24 修回日期:2022-12-27 出版日期:2023-09-28 发布日期:2023-09-27
  • 通讯作者: 张晨牧 cmzhang@ipe.ac.cn
  • 基金资助:
    国家自然科学基金资助项目;国家重点研发计划资助项目

Analysis of influence mechanism of zinc volatilization rate based on neural network prediction

Zhi ZAN1,2,  Chenmu ZHANG2*,  Jijun WU1*,  Yao SHI2,  Langming LIU3,  Weiping LIU3,   Caibei ZHUANG3   

  1. 1. Faculty of Metallurgical and Energy Engineering, Kunming University of Science and Technology, Kunming, Yunnan 650093, China 2. CAS Key Laboratory of Green Process and Engineering, National Engineering Research Center of Green Recycling for Strategic Metal Resources, Institute of Process Engineering, Chinese Academy of Sciences, Beijing 100190, China 3. Zhuzhou Smelter Group Company Limited, Zhuzhou, Hunan 412007, China
  • Received:2022-10-24 Revised:2022-12-27 Online:2023-09-28 Published:2023-09-27

RichHTML

PDF

摘要： 浸出渣回转窑煅烧回收锌、铟等有价金属是湿法炼锌行业资源绿色循环的关键环节，呈现多因素耦合、大时滞等特点，能耗高、锌挥发率不稳定，快速优化调控困难。以国内30万吨/年锌浸出渣回转窑煅烧工程为研究对象，在工况参数灰色关联度定量分析的基础上，引入粒子群算法优化建立BP神经网络锌挥发率预测模型，结合反应机理和单因子情景分析法，重点考察了焦粉投入强度、温度和浸出渣关键组分对锌挥发率的影响规律。结果表明，焦粉投入强度对锌挥发率影响显著，关联系数达0.842；同时，锌挥发率预测模型R2达0.987，整体误差≤±0.6%；焦粉投入强度、窑尾温度和浸出渣含Fe率最优模拟调控值分别为0.60 t/t, 680℃和23wt%。本研究可为湿法炼锌行业锌浸出渣绿色高质循环利用的优化控制提供理论指导和技术支撑。

关键词: 浸出渣资源化, 锌挥发率, 灰色关联度分析, PSO-BP神经网络, 情景分析, 优化控制

Abstract: The recovery and reuse of zinc and other valuable metals in leaching residues is a key segment in the green recycling of resources in the zinc hydrometallurgy industry. The typical process of zinc leaching residues treatment in rotary kilns is characterized by multivariate coupling, large delays, therefore, extensive energy consumption, unstable zinc volatilization rate and other problems arise, which is hard to be optimized rapidly and regulated immediately. The research object is about the recovery engineering of leaching slag in the large-scale rotary kiln of 300 000 tons/year in China. A particle swarm optimization BP neural network to predict the zinc volatilization rate had been established as a prioritization scheme in conjunction with a grey relational analysis of the main process parameters. Based on the single factor scenario analysis method, three model scenarios such as coke powder, kiln tail temperature, and mainly associated element of Fe content in the leaching slag had been set up, which were applied to analyze the trend and the impact mechanism of three aspects on zinc volatilisation rate. The results showed that the coke powder input intensity had the greatest influence on the zinc volatilisation rate and the correlation coefficient is 0.842. Meanwhile, the fit goodness of the PSO-BP (Particle Swarm Optimization Back Propagation) prediction model reached 0.987 and the prediction error is within ±0.6%, which achieved fast prediction of zinc volatilization rate and well solved the industrial process lag problem. The effect mechanism of coke powder input intensity, kiln tail temperature, and Fe content of the leaching residues on the volatility of zinc was illustrated in conjunction with the chemical reaction mechanism. Under the condition that the other influencing parameters were taken as the average of the sample data for the stable working conditions, the optimal simulation values for coke powder input intensity, kiln tail temperature, and Fe content of the leaching residues were 0.60 t/t, 680℃ and 23wt%. The theoretical guidance for the energy-efficient recovery of zinc from leaching residues and the optimal regulation of prevention and control of secondary pollution was demonstrated in the research.

Key words: resource of leaching residue, zinc volatilization rate, grey relational analysis, PSO-BP neural network, scenario analysis, optimal regulation and control