Modeling For The Prediction Of The Concentrate Grade In Bauxite Froth Flotation Process

Froth flotation is widely used in mineral separation. The concentrate grade is an important quality indicator of the froth flotation process. At present, most of the flotation plant acquire the value of concentrate grade through offline laboratory. However, offline laboratory takes long time resulting in adjustment hysteresis. And so, we study how to build an accurate model to predicit concentrate grade, and providing the basis for the establishment of the accurate optimal control of the flotation process.This paper developed a mechanism model of the concentrate grade after deeply analyzing the flotation principle and process through establishing the relation model of concentrate grade and recovery and the model of recovery separately. Though the model can well reflect the industrial process, the output of the model is not very well due to too many predigestions and assumptions, and its accuracy can not meet the industrial demand. So, further disscussed the data-driven pridiction model based on the image textures of foam. In this paper, the structures of the bubble image acquisition system was presented, and the method of how to get the flotation cell surface foam’s characteristics were also studied. After that, the bubble’s image features data was preprocessed through deleting outline data and denoising with wavelet. Accordingly, this paper built a BS-PLS model to predict concentrate grade takes image data as input. Considering that the model can meet a higher accuracy, but its ability of generalizing is not very well and its stability cann’t meet the need, so an integrated model was established by using information entropy; For the sake of ensuring the accuracy of the integrated model, adopted the moving window method to update the parameters and the weights of the integrated model, and used the output deviation to correct the output. The field data simulation results show that the integrated model’s result is better than any of the two single models, and it well meets the accuracy and stability requirements in industrial applications, so it can help us to establish optimal control system.