| 摘要: |
| 【目的】针对地下水水质预测中影响因素的非线性关系，采用非线性偏最小二乘回归技术（PLSR）模型进行地下水水质预测研究，为地下水水质的准确预测提供支持。【方法】运用拟线性方法建立非线性PLSR模型，选用核函数对原自变量进行非线性变换，以陕西咸阳市某观测井2001-2009年地下水资料为研究对象，进行地下水硬度预测，并将其与BP网络模型的预测结果进行比较。【结果】利用咸阳市地下水前8年（2001-2008）的水质资料建立非线性PLSR模型，采用该模型对咸阳市地下水2009年硬度进行预测，与实测值相比，非线性PLSR模型、BP网络模型预测结果的平均相对误差分别为0.944%和1.354%，可知非线性PLSR模型具有更高的预测精度和实用性。【结论】基于核函数变化的非线性PLSR模型，将复杂的非线性问题转化为简单的线性问题，简化了计算过程，提高了预测精度，为地下水水质的预测提供了一种新思路。 |
| 关键词: 地下水 水质预测 PLSR 核函数 高斯函数 |
| DOI： |
| 分类号: |
| 基金项目:国家科技支撑计划项目(2006BAD11B05) |
|
| Prediction of groundwater quality based on nonlinear PLSR model |
|
|
| Abstract: |
| 【Objective】In light of the nonlinear relationship of the factors of the groundwater quality prediction,the nonlinear partial least-squares regression (PLSR) model is used to predict the groundwater quality,supporting the accurate prediction of groundwater quality.【Method】Quasilinear approach is used to build nonlinear PLSR model,the kernel function is chosen to transform each dimension of the original independent variables into new variables.Using the data Shaanxi Province,from 2001 to 2009 of groundwater of Xianyang City,groundwater hardness is predicted,and the result is compared with the result of BP network model prediction.【Result】The first 8 year data of water quality of Xianyang City are used to build nonlinear PLSR model to predict the 2009 groundwater hardness of Xianyang City.Compared with the measured data,the average relative error of the prediction of nonlinear PLSR model and BP network model is 0.944% and 1.354%,so nonlinear PLSR model has higher prediction accuracy and strong practicality.【Conclusion】The nonlinear PLSR model based on kernel function can transform the complex nonlinear problems to simple linear ones,and effectively simplify the calculation and improve the prediction accuracy,and provide a new way to predict groundwater quality. |
| Key words: groundwater water quality PLSR kernel function Gaussian function |
