Application Of Improved BP Algorithm In Surface Water Quality Evaluation

Water environmental issues increasingly become important factors that affect human health, economic development and social stability. With the growing pollution of the water body, the importance and urgency in terms of environmental protection and improvement of the water environment monitoring and water quality evaluation system is also becoming increasingly prominent. Study of water quality evaluation the many fruitful evaluation methods and models, such as:single-factor evaluation method, fuzzy comprehensive evaluation method, BP neural network evaluation method. Due to the non-uniformity of the waters pollution factor, and geographical characteristics of, coupled with the incomplete nature of the model itself, these methods have certain limitations.BP neural network evaluation method of slow convergence speed in the existing problems in the process of training evaluation, in order to accelerate the learning speed, this paper uses the L-M optimization algorithm of traditional BP neural network evaluation model is improved, according to the national assessment of surface water quality standard, training test of the improved model using monitoring data monitoring of key areas in China in Taihu, and compared with BP model, the improved model can be better applied to the evaluation of water environment.The improved algorithm enhances the stability of the BP network structure, and improves the network learning efficiency and convergence speed. And comprehensive consideration of the attainability, representativeness and quantification of the indexes needed in evaluation process, and index system comprises4evaluation elements.100evaluating indexes were constructed. The network model is built by suing the MATLAB software and its neural network toolbox to implement the evaluation process.finally using8sets of simulation data were used to train the BP neural network improved learning, The test results show that the evaluation index system and the improved BP neural network model are practical in a good way. Comparison shows using the improved BP Artificial Neural Networks to evaluate surface water quality is more feasible, trustful and objective.