Regional Groundwater Dynamic Changes Tendency Based on Wavelet Neural Networks Model

摘要

In order to predict the dynamic changes of groundwater level, exploit water resources properly, the wavelet neural network model of ant colony optimization is put forward.Coupled model is reformed: network parameter is transformed to city matrix; total standard deviation is used for information renovation;Marr wavelet function is used to replace S shaped function of BP network. The coupled model transformed is used to predict groundwater. When heuristic pheromone intensity Q=100 , ant number M=60 ; wavelet neural network structure12-12-1, the net is convergent when the maximum number of training is 15, error accuracy is 0.0019,regression level is good. It provides a new method for study on regional groundwater dynamic change laws.