The Prediction of Bubble Trajectory Based On ACEMD Filter and Double-RBF

The bubble /micro-bubble have been applied in clinical medicine with outstanding achievements. The analysis and prediction of bubble trajectory are necessary to accurate control bubble /micro-bubble. Due to the uncertainty of liquid flow and the instability of bubbles, the bubble trajectory data is usually containing noise. So it is difficult to predict the bubble trajectory accurately. For better accuracy of prediction, a new Autocorrelation Empirical Mode Decomposition (ACEMD) filter method is proposed to remove the noise in the trajectory signal firstly. In ACEMD, the autocorrelation of signal and energy criterion are combined to overcome the shortages of conventional EMD. Then in order to obtain more information in data comprehensively, we design a Double Radial Basis Function (Double -RBF) model which contains the component RBFs and the modification RBF. The component RBFs are designed to predict the intrinsic mode functions (IMFs) of bubble data. The modification RBF is constructed to obtain the final prediction according to the results of component RBFs and denoised data. At last, the experiments are made for testing the proposed filter method and Double-RBF model. The results of experiments demonstrate that the prediction model based on new method has the best and satisfied performance.