离子交换法钨冶炼过程建模与控制策略研究

Tungsten hydrometallurgy process is a complicated physical and chemical change process, which is accompanied with severe uncertain various factors such as multi-interference, nonlinear, strong coupling , large delay, and so. In particular, it is more serious with the utilization of low-grade, complicated and fluctuant components tungsten mineral and waste tungsten residue. So it is difficulty to online detect and automatically control, and it affect on the stability of process and product quality. A new technical solution for this case is given in this study. Firstly, a computer aided process plan design and process parameter optimization is investigated by the case-based reasoning according to the history process plan cases and the database of process principle, production and operation. Secondly, the fast online detection and analysis of process solution component concentration is studied on the basis of the flow injection analysis, spectrophotometry, electrical conductivity, and machine vision chromatography. And the soft measurement models and their correction algorithm of a time-varying process parameters are built for the main process units of alkali boiling reaction, ion exchange and APT crystallization with the help of the modeling approach as artificial neural network and support vector machine. Thirdly, to enhance the system anti-interference ability, a multi-source data fusion and compensation algorithm is constructed by use of the soft-measured value, online analysis value and case-based reasoning optimal value of the process parameters in different time interval in accordance with the rolling optimization rule of predictive control. At last, the adaptive predictive control technology system of tungsten hydrometallurgy process is integrated by means of the connection of the computer aided process plan design and process parameter optimization with the soft measurement, online detection analysis, multi-source data fusion and compensation algorithm and model predictive control, thus which may be made the tungsten hydrometallurgy system adapt automatically to the various materials of complicated and fluctuant components tungsten mineral and waste tungsten residue, ensure the stability of the process and improve the tungsten recovery rate and the product quality.

钨矿源复杂，冶炼工况多变，影响过程性能的因素广泛，常规建模和优化方法难以解决这类多变量、强耦合和具严重不确定性的复杂非线性系统的建模与多目标优化控制问题。课题以低品位杂钨矿冶炼为研究对象，提出采用机理建模和动态补偿建模相结合的策略建立钨冶炼过程多模式智能混合模型的方法；综合分析钨冶炼过程参数和生产质量等指标对过程控制的需求，建立基于过程模拟、工艺案例结合的多目标融合约束模型；运用人工智能、软测量和在线检测与多目标优化理论，研究基于混合模型和多目标约束的钨冶炼过程模拟的离线优化、在线校正的多目标优化控制方法；建立钨冶炼工艺过程优化控制方法及仿真实验平台，验证并改进理论研究结果。课题研究对提高钨冶炼智能化水平，实现钨冶炼工艺系统自适应复杂矿源，保证过程稳定、提高钨回收率和产品质量、降低消耗具有重要现实意义。并可为具有类似特征的复杂系统建模、控制与优化提供新的思路，具有一定的理论意义和学术价值。

钨冶炼可用优质钨矿资源渐少，不得不处理成分复杂多变矿物且每批料须工艺试验定流程，使冶炼工况多变且过程调控频繁且难；钨冶炼工艺设备又在大型密闭且有的带温带压和复杂物理化学反应及腐蚀等环境中运行，致工艺过程监控的在线检测分析手段缺乏，普遍依靠人工定时取样离线化验来生产监控，生产过程难于自动化。虽已有钨冶炼生产线采用了自动控制技术，但未涉及工艺过程自动化和智能化。行业普遍存在产品质量波动大而难为高端材料开发提供优质原料的问题。因此，本项目以钨冶炼工艺过程这类多变量、强耦合和具不确定性的复杂非线性系统的自适应预测优化控制为目标，开展了基于案例与模拟结合的低品位杂钨矿冶炼工艺CAD优化设计及多目标融合约束模型研究、钨冶炼工艺机理建模与人工智能结合的工艺过程参数软测量方法与在线检测分析技术研究以及基于混合模型和多目标约束的钨冶炼过程模拟之离线优化、在线校正的多目标预测优化控制方法等应用基础研究和离子交换法钨冶炼工艺过程控制策略关键技术研究。主要成果有：实现了“将预测优化控制与离子交换法钨冶炼工艺过程特性相结合、过程参数在线检测分析与软测量模型预估相结合、优化设定与闭环控制相结合、过程建模与控制相结合”的离子交换法钨冶炼工艺过程智能优化控制策略，建立了离子交换法钨冶炼工艺流程模拟及工艺CAD、软测量与在线检测分析、预测优化控制等系列方法，开发了离子交换法钨冶炼工艺过程建模与优化控制仿真实验平台，解决了离子交换法钨冶炼工艺CAD、参数优化整定、在线检测分析与工艺过程优化控制等应用问题。.项目研究成果已在章源钨业股份公司等企业进行了不同钨矿源的钨冶炼工艺流程模拟与工艺CAD、主要单元工艺过程参数软测量与在线检测分析以及对除杂净化、APT离子交换等单元工艺关键点的检测与判定和工艺流程优化控制等试点应用，表明制定生产工艺流程的效率提高20%以上、人工取样和离线化验频次减少50%以上；工艺过程自动控制可降低人的干预和减轻操作人员的工作压力，制品质量一致性得到了显著改善。项目成果可推广应用至诸如稀土、铝等其它有色金属冶金领域。.项目研究对离子交换法钨冶炼工艺过程的先进控制和智能化提供了方法与技术基础，具有重要的理论意义；对实现钨冶炼工艺系统自适应复杂矿源、提升质量、节能降耗和提高钨冶炼智能化水平具有重要现实意义和应用价值。