The flotation froth image recognition-based concentrate ash prediction system

For realizing real-time monitoring of flotation process, a set of intelligent flotation froth image recognition systems is applied at Liuwan Mine Coal Preparation Plant. Through the froth images captured by the cameras fitted respectively on the first and the third cells of the flotation machine, the features of color, collapse rate, size, form and flow velocity of bubbles can be extracted. Based on the features extracted, a concentrate ash prediction model is developed using support vector regression algorithm. Practice shows that compared to the actually measured ash values of the concentrate respectively produced in the first and the third cells, the absolve errors of the predicted values are 0.34% and 0.37% respectively.