摘要
河道水质风险分析是一个涉及参数较多、而数据资料较少的复杂统计问题。将水质监测值看作水质指标分布的一个样本,结合水质指标分布参数及超参数先验分布,在推导各参数及超参数满条件概率公式基础上,引入MCMC理论中的Gibbs抽样方法,建立了基于MCMC理论的河道水质风险分析模型。通过实例验证了模型的合理性,为水质风险分析提供了新的思路和方法。
River water quality risk analysis is a complicated statistical problem involving multi-parameters but insufficient data available.This study considers each set of water quality monitoring results as a sample of water quality distribution indices,and derives a full condition distribution for each parameter and super-parameter.Then by assigning a prior distribution to each of these parameters,a water quality risk model is constructed,and this multi-parameters problem is solved using Gibbs sampling method in the Markov chain Monte Carlo（MCMC） theory.Sufficient samples has been drawn to evaluate the statistics of each parameter.Case studies shows the rationality of this model that provides a new method for river water quality risk analysis.
出处
《水力发电学报》
EI
CSCD
北大核心
2012年第6期77-82,共6页
Journal of Hydroelectric Engineering
基金
国家教育部博士学科点专项科研基金(20100141120029)
水利部公益性行业科研专项项目(201001003-6)
中央高校基本科研业务费专项资金资助
关键词
环境水利
水质
风险
MCMC
抽样
enviroment hydroulic
water quality
risk
MCMC
sampling
