Gray RBF Network Modeling And Application In Eutrophication Prediction Of West Lake, Hangzhou

Scientific technology developed rapidly recent years, so did computer and artificial intelligence technology. Artificial neural network, the production of fast development of computer and artificial intelligence technology, has been applied in complex systems to deal with problems of approximate modeling, because of its powerful adaptability.The study selected Back Propagation (BP) and Radial Basis Function (RBF) neural network, which were widely applied in network research neural network (NN) to investigate data about West Lake water quality measured during 2000-2001. Gray relationship analyse and gray theory were used to pre-treat the initial data and then selected presented water quality parameters as the variable to establish the network model. Water temperature, pH as the input variable, chlorophyll-a as the output variable founded network and compared forecast precision, network convergence velocity among the three networks. Results showed that gray RBF neural network connected with gray theory appeared better capability in data fit. The error between measured water quality data about West-lake and short-time forecasted modeling data was narrow, which showed that gray RBF neural network was capable to simulate water quality change trend, and could provide scientific guidance to water eutrophication forecast and treatment.