基于机器视觉的镨/钕分离过程组分含量在线监测方法研究

The rare-earth elements are known as " industrial vitamins ". At home and abroad, solvent extraction is a key technology which is widely used to obtain the high-pure and single rare-earth element. And the continuous, fast and reliable monitoring of their component content is a very difficult problem, which needs to be solved in the rare-earth extraction separation process.This subject aims at the Pr/Nd extraction separation system with the ion characteristic color. And machine vision technology is applied to dynamically identify the distribution change of ion characteristics color in the extraction tanks, so the deviation of change of the elements component content can be monitored on-line. For this purpose, a video information acquisition and image processing system for the rare-earth mixed solution color is built. Different feature subspace models of rare-earth ion color image are discussed, and an extraction method of ion color feature, which has a strong adaptability to the change of external environment, is proposed. Then, a detection model of element component content based on the image characteristics, which is of good robustness and generalization, is developed by intelligent integrated modeling method. To validate and refine the theoretical results with field test, we also design and develop an online-monitoring system of element component content based on machine vision. The research has a vital significance for enriching and perfecting online-monitoring methods in the extraction process, and supporting the optimal operation of the extraction separation with the ion characteristic color. Simultaneously, it provides a new idea for the key parameters detection of complex industrial processes with analogous feature, and has a certain theoretical significance and application prospect.

稀土元素享有"工业维生素"之称。溶剂萃取法是国内外普遍采用的获取高纯单一稀土的主要工艺，萃取过程中元素组分含量连续、快速、可靠检测一直是行业亟待解决的难题。课题针对具有离子特征颜色的Pr/Nd萃取分离体系，采用机器视觉技术，通过实时识别萃取槽体中离子特征颜色分布变化动态监测稀土元素组分含量变化。课题拟构建稀土混合溶液颜色视频信息采集和图像处理系统，探讨稀土离子颜色图像的不同特征空间模型，研究提出对外界环境变化具有较强适应性的离子颜色特征提取方法；采用智能集成建模方法，建立基于离子颜色特征、具有较强鲁棒性和泛化能力的元素组分含量检测模型；开发基于机器视觉的元素组分含量在线检测系统，进行现场试验验证，改进和完善研究成果。课题研究将完善稀土萃取过程在线检测手段、为具有离子特征颜色的稀土萃取过程优化运行提供支撑，同时为具有类似特征的复杂工业过程关键参数检测提供新思路，具有一定的理论意义和应用前景。

稀土元素享有“工业维生素”之称。溶剂萃取法是国内外普遍采用的获取高纯单一稀土的主要工艺，萃取过程中元素组分含量连续、快速、可靠检测一直是行业亟待解决的难题。课题针对具有离子特征颜色的Pr/Nd 萃取分离体系，采用机器视觉技术，通过实时识别萃取槽体中离子特征颜色分布变化动态监测稀土元素组分含量变化。课题构建了稀土混合溶液颜色视频信息采集和图像处理系统，探讨了稀土离子颜色图像的不同特征空间模型，研究提出对外界环境变化具有较强适应性的离子颜色特征提取方法；采用智能建模方法,建立了基于单一颜色空间和多颜色空间融合的稀土萃取过程组分含量预测模型，开展了组分含量模型自适应校正方法研究；开发了基于机器视觉的元素组分含量检测系统，将所提方法进行了相关的试验验证。课题研究完善了稀土萃取过程检测手段、为具有离子特征颜色的稀土萃取过程优化运行提供支撑，同时为具有类似特征的复杂工业过程关键参数检测提供新思路，具有一定的理论意义和应用前景。.研究过程中申请国家发明专利2项；发表研究论文28篇，其中SCI收录3篇、EI收录6篇，培养毕业博士研究生1人、毕业硕士研究生8人、在读3人。完成预期目标。