Pattern Recognition of the Soft Abrasive Flow Based on Wavelet Packet

A Pattern Recognition method based on wavelet packet and artificial neural is proposed for solid-liquid two-phase flow characteristic parameters and the non-linear relationship between flow pattern. This method firstly established the physical and dynamic model, then set a monitoring point. To get the optimum wavelet tree and its information entropy, six floors of wavelet packet was used to decompose the collected velocity fluctuation signal. Transported the proper vector which is component by information entropy into Back Propagation neural network to train and identify. The recognition results show that this method can effectively overcome the subjectivity of traditional identification methods. It has good recognition effect, thus provide an effective choice for solid-liquid two-phase flow pattern recognition.