神经网络在次级河流回水区叶绿素a浓度预测中应用_罗固源.pdfVIP

3 2 环境 工程学 报 Vol. 3, No. 2 2 0 0 9 2 Ch inese Journal of Environm entalEngineering Feb . 2 0 0 9 a 1 1* 1 2 2 罗固源 郑剑锋 许晓毅 曹 佳 舒为群 ( 1. , 000 5; 2. , 00038) , a (PCA) a, RBF , , RBF a a, a TP183; X 171. 1 A 1673-9108( 2009) 02-0372-05 Application of theneuralnetwork in prediction for chlorophyll-a in branch backwater region 1 1 1 2 2 Luo Guyuan Zheng Jianfeng Xu X iaoyi Cao Jia Shu W eiqun ( 1. Key Laboratory of the Three GorgesReservoir Reg ion. s E co-Environm ent, M inistry ofE ducation, Chongqing University, Chongqing 000 5; 2. Faculty ofM ilitary PreventiveM edicine, ThirdM ilitaryM edicalUniversity, Chongqing 00038) Abstract Tak ing L injiang riverw hich is a branch ofY angtzeR iver as research object to evaluate the feas-i b ility of neuralnetwork model for smi u lating chlorophyll-a trend in branch backw ater region. By using themethod of principal component analysis ( PCA ) to select themain indexeswh ich affect the chlorophyll-a trend, theRBF neural netw ork modelw as created based on the database of indexes. The training and testing resu lts ofmodel in- dicated that the smi u lating accuracy ofmodelw as h igh; it show ed that the RBF neural network model could be used for smi ulating the chlorophyl-l a short-term trend in branch backw ater region. By analyzing the influencing factors of chlorophyll-a in L injiang river backw ater region, the resu lt show ed that controlling phosphorus content w ou ld be mi portant to pre