Research On Prediction Method Of Bubble Rising Trajectory In Water

Mineral flotation is one of the methods of mineral extraction widely used recently,which can increase the mineral utilization rate by changing the difference of the surface wet-ability of ore grain to distinguish the useful and useless minerals.However,due to the uncertainty of gas and liquid movement,the bubble trajectory in flotation process is too complicated to control accurately in practical application,which brings not only operation difficulties,but also waste of resources as well.Treating the bubble's trajectory in the basic flotation unit as research object,this dissertation proposed a new model of prediction after denoising.A new denoising method combining the improved energy denoising algorithm with the empirical modal decomposition(EMD)algorithm is proposed for the trajectory data of air bubble movement with irregular noise.This method defines noise range in mathematical mode,making up for the lack of clear definition of noise in traditional methods.Then,the commonly used neural network algorithm is introduced.An extremum correction algorithm based on RBF neural network is proposed for the end effect in EMD.With the self-correlation of signal,this algorithm is used to revise the RBF prediction results of high precision.The improved energy denoising algorithm is used to eliminate the random noise signal in bubble's rising trajectory data,extending the two endpoints of the signal and then the complete signal is obtained.The complete signal is decomposed by EMD to its original region,and multiple RBF prediction models are then established.The final results are obtained by summing the prediction results of each model.Compared with the traditional algorithm,it is showed that present method has significant improvement in both denoising and prediction.The bubble trajectory curve is smoother and the prediction result is more accurate,so this model can provide necessary data support for the analysis of flotation bubble motion.