The Research Of The Prediction Of Flotation Recovery Based On Improved LS-SVM

In the process of anionic reverse flotation, flotation recovery is a very important economic indicators, It is the most important criteria Peugeot of technology and management level in a concentrator, so is necessary for accurate prediction of recovery in order to better control. But because of the strong non-linearity and uncertainty between the reagent and its influence factor, it is difficult to be described with accurate mathematical models, for a long time, it has restricted people to make accurate predictions.To solve the foregoing problem, a new recovery prediction method is proposed in this paper based on the actual production of a flotation concentra- tor. The prediction model is built by a improved weight LS-SVM with image features as input,and the regulating parameters are optimized by GA.First of all, the paper extracted the characteristics of the field of foam image, dug out the relative redness component, bubble size, bubble load, velocity and bubble collapse and flotation process flotation recovery exist certain corresponding relation.After above, data in the paper is classified and disposed by FCM, and is cleaned based on Genetic Algorithm and Neural Network to clear the data which contains noise, incomplete and inconsistent.Then, researched the relationship between characteristics of the field of foam image and the flotation recover further, and established the prediction model based on the relationship between variants. The prediction model was built and simulated in MATLAB. Results show that the prediction model established through improved weight LS-SVM can forecast the flotation recovery in the process of flotation more accurate, and achieve the need of production process.Finally, this paper designed the process of flotation flotation recovery prediction software modules. The software module implemented the visual query of the data and the prediction of flotation recovery. The software module was mainly based on the VB.net and MATLAB platform. Vb.net mainly developed the visual interface module and some algorithms design, MATLAB mainly developed the other algorithms design such as online trends which can be called by VB.net to completed two complementary advantages of the programming language.