Prediction Of Eutrophication By BP Neural Network Model Based On PSO Algorithm

Rivers, lakes, reservoirs'eutrophication prediction has became a hot research by scholars at home and abroad, because the water body ecological environment is complicated and diversified, the environmental effect factors are numerous, the traditional experience forecast model already is not suitable highly the misalignment water quality forecast.therefore, a model of the BP neural network which based on the particle swarm optimization (PSO) algorithm is proposed and its application on nonlinear prediction of eutrophication in Ming Lake is also discussed further, which has a certain scientific significance in the development of water evaluation and prediction.In prediction of eutrophication by BP neural network, the fifth sampling point which can most representative of Ming Lake water quality was taken as the research object, chlorophyll a is the characterization of the phenomenon and the degree of eutrophication refers to the most important indicator of the output variables to determine the network; monthly samples collected from April 2009 to March 2010 from Ming Lake were interpolated as weekly samples(training regulations), by the spline function of MATLAB; network input variables was screened by the correlation analysis; determining the 4-layer network structure with the concealed level node 12. Short-term variation trends of the fifth sampling point's chlorophyll a concentration were predicted by BP network which completed using the training regulations trained. Results showed that, prediction of BP neural network to the fifth sampling point has a good network of test performance, however, the traditional BP algorithm with limited optimization of the algorithm system has slow convergence.Based on the limitation of traditional BP network, in the PSO-BP neural network prediction of eutrophication, the structure of the neural network and the overall convergence of the algorithm were optimized according to the adjusted weight value of the neural network by utilizing the PSO algorithm, a model of the BP neural network based on the PSO algorithm was proposed. Short-term variation trends of the fifth sampling point's chlorophyll a concentration were predicted by BP-PSO network which completed using the training regulations trained. Results showed that, PSO algorithm effectively overcame the shortcomings of the traditional algorithm and PSO-BP network model shows higher test performance at the fifth point.Learning and generalization ability of the BP neural network were verified using data from the sixth and the first sampling point. Comparing the prediction accuracy of the BP neural network based on the PSO algorithm with the traditional BP algorithm. Results showed that, the new model on the sixth point and the first point of the relative error about 3% and 6%, was lower than the traditional BP model 8% and 10%. It is proved that the PSO-BP neural network model for Ming Lake overall eutrophication in short-term trend has higher prediction ability and make the foundation for the PSO-BP neural network model in prediction of eutrophication in future.