基于非线性PLSR模型的地下水水质预测.pdfVIP

39 7 ( ) Vol. 39 No. 7 20 11 7 Journal of Northwest A F University( Nat. S i. Ed. ) Jul. 2011 * PLSR 郝 健, 刘俊民, 张殷钦 ( , 712100) [ ] , ( PLSR) , PLSR , , 2001- 2009 , , BP 8 ( 200 1- 2008) PLSR , 2009 , , PLSR BP 0. 944% 1. 354% , PLSR PLSR , , , , [ ] ; ; PLSR; ; [ ] P64 1. 12 [ ] A [ ] 167 19387( 20 11) 0702 1205 Predict ion of groundwater quality based on nonlinear PLSR model HA O Jian, LIU Junmin, ZHA NG Yinqin ( Colleg e of Water Resour ces and A r chitectural E ngineer ing , N or thw est A F Univ ersity , Yangling , S haan i 712100, China) Abstract: Obje tiveIn light of the nonlinear relationship of the fa tors of the groundw ater quality predi tion, the nonlinear partial leastsquares regression ( PLSR) model is used to predi t the gr oundw ater quality, supporting the a urate predi tion of groundw ater quality. M ethod Quasilinear appr oa h is used to build nonlinear PLSR model, the kernel fun tion is hosen to transform ea h dimension of the original in dependent variables into new var iables. U sing the data Shaanx i Provin e, from 2001 to 2009 of gr oundw ater of Xianyang City, groundw ater hardness is predi ted, and the result is ompared w ith the result of BP net w ork model predi tion.ResultT he first 8 y ear data of w ater quality of Xianyang City ar e used to build nonlinear PLSR model to predi t the 2009 groundw ater hardness of Xiany ang City. Compar ed w ith the measured data, the average relative error of the predi tion of nonlinear PLSR model and BP netw ork model is 0. 944% and 1. 354% , so nonlinear PLSR model has higher pr edi tion a ura y and strong pra ti ality. Con lusion The n