摘要
酸雨pH值受酸性离子[SO2-4]、[NO-3]和碱性离子[Ca2+]、[NH+4]等的影响。这些影响因素之间存在多重相关性。用一般最小二乘回归法建模预测pH值,估计参数存在着很大的误差且物理意义明显不足。应用偏最小二乘回归技术建立pH值预测模型,克服了自变量之间的多重相关性的问题,因而更具有先进性,计算结果更为可靠。以我国17个城市pH值预测为例,探讨偏最小二乘法的优势,并与最小二乘回归法进行了比较。
pH values of acid precipitation are affected by not only acid ionsand,but also alkaline ionsand .There are some multiple correlations between these factors.As the multiple correlation is existed,the estimated regressive parameters with the least square method include a good deal of errors in the multiple regressive equation which cannot reflect its physical meaning.The partial least square method can easily solve the multiple correlation problem.The method is simple and quick to calculate.The estimated regressive parameters from it are robustness.A case study,the pH values prediction of precipitation of 17 cities in China,has been researched.The results show that the partial least square regressive is better than the former.
出处
《四川环境》
2003年第6期77-80,共4页
Sichuan Environment
基金
国家自然科学基金(50279023)资助
关键词
多元线性回归
最小二乘
偏最小二乘
城市降水pH预测值
Multiple linear regressive model
least square
partial least square
pH value prediction of urban precipitation
