基于逐层试验设计的多维不确定参数对水资源配置结果的影响研究

Research on the Effect of Multidimensional Uncertain Parameters on Water Resources Allocation Results based on Layer-by-layer Experimental Design

不确定参数对水资源配置结果有着重要影响,相关研究较少定量评估不确定参数输入带来的系统响应情况变化。以区间数表示水资源配置涉及的不确定参数,构建了含多维不确定参数的不精确两阶段随机规划模型,然后采用逐层试验设计方法,量化不确定参数对目标值、配置决策等配置结果关键边界特征(上界、中值、下界)的主效应、交互效应和贡献率,最后以南渡江流域为实例验证本方法的实用性。结果表明,在规划年2030年,南渡江流域管理者可以按照预先定义的供水目标上限供水,预期供水总收益为[213.11,305.89]亿元。逐层试验设计能够定量分析多维不确定参数对配置结果的影响,中游、下游的农业和市政供水效益系数对供水总效益下界、中值、上界的贡献率可达60%以上,中游农业供水效益系数变化造成的供水总收益变化最大,对供水总效益的下界、中值、上界的主效应分别为44.22、48.28、43.01;P=95%特别枯水年来水量对总缺水量下界、中值、上界的贡献率可达50%以上,其主效应分别为-11.04、-22.30、-24.12,其变化造成的总缺水量变化最大;供水总效益或总缺水量下界、中值和上界的主要影响参数类型一致,但是不同界限上贡献率值不同,且交互效应影响很低。研究结果可为南渡江流域水资源配置方案制定和水安全保障分析提供科学依据。

Uncertain parameters exert a significant influence on the outcomes of water resource allocation.However,there remains a dearth of quantitative assessment regarding the variations in system responses stemming from these uncertain parameter inputs.This study employed interval numbers to represent uncertainties within water resource allocation,constructing an inexact two-stage stochastic programming model with multidimensional uncertain parameters,employing a layer-by-layer experimental design approach,which systematically quantifies the main effects,interaction effects,and contribution rates of uncertain parameters on critical boundary features of allocation outcomes,includ-ing upper bounds,median values,and lower bounds of target values and allocation decisions.The practical application of this methodology was demonstrated through a case study of the Nandu River Basin.Results revealed that by the year 2030,managers in the Nandu River Basin could supply water up to a predetermined upper limit,resulting in an anticipated total water supply revenue ranging from 21.311 billion to 30.589 billion yuan.The layer-by-layer experimental design facilitated a quantitative analysis of the influence of multidimensional uncertain parameters on allocation outcomes.Notably,the benefit coefficients of agricultural and municipal water supply in the middle and lower reach-es contributed more than 60%to the lower bound,median,and upper bound of total water supply benefits.Variations in the benifical coeffi-cient of agricultural water supply in the middle reaches had the most significant impact on total water supply revenue,with main effects on the lower bound,median,and upper bound of total water supply benefits being 44.22,48.28,and 43.01,respectively.Additionally,the contribution rates of water inflow at a 95%probability level during particularly dry years to the lower bound,median,and upper bound of to-tal water shortage exceed 50%,with main effects of-11.04,-22.30,and-24.12,respectively,resulting in the most substantial changes in total water shortage.The primary influencing parameter types for total water supply benefits or total water shortage across the lower bound,median,and upper bound remained consistent,although the contribution rates varied across different boundaries,with a low impact of inter-action effects.These research findings have significant implications for informing the formulation of water resource allocation schemes and an-alyzing water security assurance in the Nandu River Basin.