基于机器视觉的矿物浮选过程监控技术研究进展  

Machine-vision-based Online Measuring and Controlling Technologies for Mineral Flotation — A Review

作　　者：桂卫华[1] 阳春华[1] 徐德刚[1] 卢明[1] 谢永芳[1] 

机构地区：[1]中南大学信息科学与工程学院,长沙410083

出　　处：《自动化学报》2013年第11期1879-1888,共10页Acta Automatica Sinica

基　　金：国家创新研究群体科学基金项目(61321003);国家自然科学基金(61134006;61025015;61074117);国家科技支撑计划(2012BAK09B00;2012BAF03B05)资助~~

摘　　要：矿物浮选流程长、分布范围广、控制变量多、关键工艺参数无法在线检测,导致实时监控困难,严重制约了浮选生产的优化运行及选矿自动化水平的提升.浮选泡沫表面视觉特征是浮选工况和工艺指标的直接指示器,为此将机器视觉应用到矿物浮选过程的监控中,以提高浮选过程的资源回收率.本文结合矿物浮选泡沫图像特点,从浮选过程的泡沫图像关键特征提取及表征、关键工艺参数检测、工况识别以及基于机器视觉监控系统的实现等方面综述了浮选过程监控技术的研究成果,并指出了基于机器视觉的选矿过程监控技术的发展趋势及面临的挑战.The real-time monitoring and control of mineral flotation process are difficult due to several facts/system characteristics： long flotation processes, wide distribution range, multiple-variable control system and undetectability of crucial production parameters. All these facts/system characteristics have greatly restricted the optimal operation and automation level of flotation process. However, the visual features of flotation froth surface play the indication role to illustrate the production states and degree. As a consequence, machine vision technology is employed to facilitate the control strategy design for mineral flotation process and to promote the resource recovery. This paper reviews the key technologies and the corresponding achievements associated to the system design of the flotation process monitoring and control. Concretely, those key technologies involve the skills to extract and characterize the key froth image features, detect and identify the process parameters and production states. Moreover, the realizations of froth image based mineral flotation process monitoring system are discussed. Finally, recommendations for future research encountered in the control strategy design for flotation process based on machine vision are suggested.

关 键 词：机器视觉 泡沫浮选 在线检测 特征选择 工况识别 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TD928.9[自动化与计算机技术—计算机科学与技术]