Design And Implementation Of Recognition System On The Froth Flotation Image Of Ilmenite

Titanium resource is an important strategic resource.To efficiently extract ilmenite from the tailings of vanadium titano-magnetite,promote the recovery rate of ilmenite,using flotation automatic control system can reduce the labor intensity and optimize operating environment.Image recognition system of ilmenite foam is basis for flotation automatic control system.With control parameters provided by the image recognition system,the automatic control system of flotation can dynamically adjust the relevant parameters based on foam flotation conditions,and achieve stable and efficient production as the model is continuously optimized.Based on the flotation process of other minerals,a suit of foam image recognition system adaptable for ilmenite flotation is designed in this paper.Through the industrial camera,the foam image is captured and transmitted to the server in which pre-processing of image are carried out.By simulating quality criteria of engineers on site,important characteristic parameters such as color characteristics of the image HSI color space,the size and number of foam,foam movement speed,and the Tamura texture feature of foam are selected.These characteristics can reflect the quality and defects of foam well.Based on extracted foam characteristic parameters and acquired parameters on-site,a sample database is established with expert experiences and later test results.Due to small quantity of the sample data,combined with characteristics of the sample and the classification algorithm,the paper tests two mainstream algorithms of classification.The machine learning algorithm based on SVM finally be chosen,and an expert system based on the algorithm is established to judge the quality of foam,finally the system is developed.Judged by the expert system with real-time data,it predicts the quality and then gives alarm information or operation suggestions,output control parameters,real-time trend graphs and historical data query.The system,deployed and tested in 2# concentration tank of flotation workshop of Xinbaima Mining Co.,Ltd.,Panzhihua Mining Group,can judges foam quality close to the artificial normal judgment level,which achieves the operation guidance of flotation process and output of control parameters.The next step is going to improve the prediction level and promote stable performance of the system,and then extend to all the flotation tanks.