Based On Image Processing Of Flotation Foam Modeling Of Concentrate Grade

Foam flotation,as a method of changing the physical and chemical properties of mineral surfaces by flotation reagents to separate minerals of different compositions,is of great significance to industrial beneficiation.This thesis is based on the practical application of the project.The main research content includes the construction of the software and hardware platform of the flotation foam image acquisition and analysis system,the research of the digital image processing algorithm for flotation foam,and the construction of the concentrate grade prediction model based on GA-RNN mold.(1)The construction of the software and hardware platform of the flotation foam image acquisition and analysis system: at the site of the flotation workshop of the Jinchuan Nickel Mine,the hardware construction and software operation of the system are completed and delivered to the other party as actual engineering results.According to the characteristics of the nickel ore flotation process and the actual needs of on-site workers,the software functions are continuously improved,and four subsystems to realize the software functions are proposed,namely the image acquisition subsystem,image feature value extraction subsystem,image and data display subsystem,Operation advice and ledger recording subsystem.The software runs on the industrial computer,in the window10 operating system environment,with the help of the VS2017 development environment,using C ++ 11 language to write code and MFC framework to complete the development of desktop applications.(2)Research on image processing algorithm of flotation foam: research on image processing algorithm suitable for flotation foam,using RGB color model algorithm and HSV color model algorithm to extract color features;using color image grayscale algorithm to convert color image of flotation foam For the grayscale image,an image enhancement algorithm based on histogram equalization and an image filter algorithm based on cascaded filters are used to complete the image preprocessing work;the marker-based watershed segmentation algorithm is optimized to complete the flotation foam image segmentation,thus Extract the morphological characteristics of the flotation foam;propose the use of the ORB algorithm for feature point detection and description,the FLANN algorithm for feature point matching,and the RANSAC algorithm for error matching and filtering to extract the velocity characteristics of the flotation foam.(3)Modeling of the concentrate grade prediction model based on GA-RNN: According to the characteristics of the nickel ore flotation process,it is known that the characteristic parameters of the flotation foam image have time series characteristics,and it is determined to build a concentrate grade prediction model based on the cyclic neural network algorithm and build Preliminary model.In this paper,in view of the low prediction accuracy of the preliminary model and the shortcomings of the neural network’s sensitivity to the initial weights and thresholds,a genetic algorithm is proposed to find the initial weights and thresholds of the recurrent neural network,and the prediction accuracy of the model is further optimized.