On The Modeling Of Water Quality Evaluation By Using Improved Wavelet Neural Network

Water is a kind of irreplaceable resources.In recent years, it has always attached great importance to the work of water resources protection and management both in China and abroad.However, with the development of science and industry, the problems of water resources is always restricts the development of society and ecological environment.At the same time, the traditional evaluation method of water quality is lack of efficient processing efficiency because of the complexity and nonlinear of water environment, therefore, it is very urgent to improve the protection of water resource.The development of artificial neural network(ANN) brought a new direction for the research of water quality, both in China and abroad have a lot of research on water quality based on artificial neural network.Based on the work of predecessors in artificial Neural Network and the water quality evaluation, further study the theory, the structure and the algorithm of Wavelet Neural Network, try using Wavelet Neural Network(Wavelet Neural Network, the WNN) applied to the research of water quality evaluation.The main research in this paper includes the following several aspects:1. In view of the traditional water quality evaluation method has certain limitation, using the better convergence speed、generalization ability、precision and nonlinear processing ability of wavelet neural network, puts forward adopting wavelet neural network for water quality evaluation modeling, Comparing the experimental results of the evaluation results and the traditional evaluation results, prove the feasibility of this idea.2. Because of the slow convergence speed of the traditional wavelet neural network algorithm, so introducing the adaptive learning and momentum factor, to speed up the network learning, improved the learning ability of the network.3. Because of the algorithm of traditional wavelet neural network easy to fall into local minimum, introducing genetic algorithm(GA) in the wavelet neural network.Although the genetic algorithm has good adaptive learning ability and the global search ability, but its convergence speed is slowly, thus put forward an improved genetic algorithm-adaptive genetic algorithm(AGA) apply to optimized the neural network; At first adopts the adaptive genetic algorithm to optimize initial weights、threshold 、scaling and translation parameters of WNN, and then choose a good parameter as to improve the initial parameters of WNN, When creating a water quality evaluation model.This method combines the AGA algorithm’s globalsearch ability and local search ability of adaptive momentum gradient descent method,comparing the simulation results, prove the realizability of this theory.4. By the traditional WNN algorithm,the improved WNN algorithm and AGA algorithm build water quality evaluation model based on wavelet neural network,the simulation experiment, comparison and analysis of experiment results.The results show that model of the AGA algorithm base on wavelet neural network is better than other methods,the method to the evaluation of water environment have higher accuracy and effectiveness. At last,create a graphical user interface(GUI) for water quality evaluation based on WNN, and facilitate the user’s usage.