Research On Feature Extraction And Operating Condition Identification Of Froth Image In Copper Flotation Process

With the rapid development of the national economy,the demand for copper in China has been increasing.As an important mineral processing technology,flotation is widely used in the production of copper.However,flotation production automation level is low in China.The subjectivity and randomness of artificial observation led to the unstable operation of flotation.Flotation production is less efficient.The waste of mineral resources is serious.Therefore,the accurate identification of flotation conditions is the key to the optimal control of flotation process,which has important practical significance.During flotation,the apparent characteristics of froth are closely related to working conditions.The image processing based flotation froth condition Identification technology has attracted extensive attention from experts and scholars because it can accurately describe the image features and completely overcome the subjectivity and randomness of manual observation.Copper flotation is the research background of this paper.Through the feature extraction of copper flotation froth image and the establishment of flotation condition identification model,the accurate identification of flotation conditions is realized.The main research contents of this paper are as follows.1.In order to dig out the information between the similar gray level and multi-scale characteristics of froth images,a texture feature extraction method based on symbiotic augmented matrix in dual-tree complex wavelet domain is proposed in this paper.The simulation proves that the texture feature extraction algorithm is not only time-consuming,but also effective and recognizable.2.The image of flotation froth is more noise and highlights adhesion.The uneven distribution of bubble size can lead to over segmentation and under segmentation.To solve these problems,a watershed segmentation algorithm based on improved gray range transform is proposed.Based on the algorithm,the flotation bubble size feature extraction has strong recognizability and anti-interference.3.Through the research on the identification method of flotation condition,a multi-core support vector machine(MKL-SVM)condition recognition method based on improved decision binary tree is proposed.Based on this method,a flotation condition recognition model is established.The extracted flotation froth texture feature and size feature is used as the input feature vector of the model to train and test the model.The results show that the flotation condition identification method proposed in this paper has high recognition rate and high antiinterference,and has high application value.