Research And Implementation Of Flotation Process Monitoring System Based On Froth Image

Froth flotation is a widely used process to extract certain minerals from ore. Since a long time, in order to implement the process control of flotation, almost all the factories depend on human observation to obtain physical characteristics of bubble surface. However, owing to frequent personnel turnover, unreliability or inaccuracy of manual manipulation, medicament accession may become confused. Consequently, the process control of flotation can't be guaranteed to implement optimally. So, a kind of froth image based flotation process monitoring system was researched and developed in this paper, thus to provide guidance for optimization of flotation process, cut resourse comsumption and keep competitive power of enterprises.Firstly, the properties of froth images were summurized, then the interrelationship between bubble characteristics and process indicators was analysed, and also the crucial image processing algorithms required for system realization were researched on.Secondly, requirement analysis and schematic design of system were presented. Trough comparing among types of froths generated by different flotation cells, in conjunction with analysis of concentrate quality to which contributed by each flotation procedure, cameras' installing spots were finally determined. Meanwhile, according to workers'actual demands and flotation's specifities, system's fundamental function and real-time capabilities were pointed out. Based on this, design schemes of system were proposed for handware and software respectively, each subsystem structure was presented and the dynamic collaboration relationships of main use cases were described. Moreover, software dababase design scheme was given.At last, realization of system's key technique was given prominence in this paper. Extraction of bubble velocity and shape is such a focal spot that a macro-block-tracking combined with three-step-search based algorithm and a maximum potential energy based watershed algorithm were proposed to extract velocity and shape characteristics respectively, of which the detail realization procedures were also described. Then, a multi-threads design scheme was proposed to overcome synchronization problems among threads. Furthermore, according to difficulties of image acquisition and parallel execution of multi-cameras, a solution idea and the conrresponding realization procedure were expounded. Optimal control strategy for flotation process based on expertise knowledge system was exhibited additionally.The system developed synthesizes lots of functions, including image acquisition, feature extraction, trend graph plotting, data visualization and storage, parameters prediction, expert guidance of operation suggestions and so on. It is verified that this system has strong versatility and wide range of applications to meet the diverse needs of video surveillance occasions.