Establishment Of BOD-DO Water Quality Simulant Model Coupling With BP Neural Network And Study On Its Application

Artificial Neural Network (ANN) is the forefront of complex non-linear science and artificial intelligence science. In recent years, with increasingly perfect of ANN theory and applied technology, ANN is gradually becoming the focal point of every subject. Based on general analysis of water quality simulant model, basic principle of ANN and optimized arithmetic, the paper suggested introduce Back—Propagation Network(BP network) into the field of water quality simulation and water quality forecast. The paper did some explorative research in Artificial Neural Network modeling methods in water quality simulation and its adaptability in water quality forecast, and made efforts in improving the level of water pollution control programming.Based on the analysis of subsection transfer character of pollution in the course of transfer in river, the paper put forward series-wound ANN model, which simulation transfer rule of pollutant in the river. Based on this modeling idea, we upbuilt BOD-DO coupling BP network water quality simulant model, adopted the model of one-dimensional water quality synthetically simulation BP network ,and did training to the model with neural network toolbox in Matlab. Our research manifested: the result of ANN model simulating water quality has higher precision than the result of one-dimensional water quality model, validated the availability and exactitude of the structure of series-wound ANN model.The superiority of BOD-DO coupling BP network water quality simulant model is that it needn't complicated mathematical modeling, could adapt complicated and even unexpectable changes of condition, and what's more, its precision is higher than mathematical modeling, it is specially adaptive to the establishment of multiplex non-linear model.