摘要
【目的】年降水量预测对揭示区域水资源动态变化特性具有重要意义。为提高集对分析预测模型精度,开展基于集对分析的年降水量预测模型及应用研究。【方法】以淮北地区艾亭站1956—2003年降水量为历史样本,利用自相关分析确定历史集合和当前集合,以均值离差法、距平百分率法、均值标准差法确定等级标准及联系数;采用特殊值法、减法集对势法和半偏减法集对势法计算集合间联系数值;依据集对同势原则识别相似历史集合,经算术平均法和加权平均法确定预测值,进而构建基于集对分析的年降水量预测模型,并应用于2004—2008年的降水量预测,提出适用于集对分析预测模型构建的最优组合方法。【结果】以同势原则确定历史集合是合理的;采用算术平均法计算预测值的精度是可接受的;减法集对势和半偏减法集对势确定联系数值相较于特殊值法更优;3种等级标准划分方法的预测误差均在30.00%以内,均值离差法的平均相对误差为11.58%,距平百分率的为14.01%,均值标准差法的为13.74%。【结论】将集对分析法用于年降水量预测是可行的;减法集对势和半偏减法集对势兼顾了物理特性和动态变化属性,均值离差法综合考虑了数据的集中趋势和离散程度,更适用于基于集对分析的预测模型构建。
【Objective】The prediction of annual precipitation is of great significance in revealing the dynamic change characteristics of regional water resources.To improve the accuracy of the set pair analysis prediction model,a study on the annual precipitation prediction model and its application based on set pair analysis is carried out.【Methods】The precipitation from 1956 to 2003 at Aiting station in Huaibei area was taken as a historical sample.Firstly,the historical set and current set were identified by autocorrelation analysis,and the grading criteria and the connection number were determined by the mean deviation method,the anomaly percentage method,and the mean standard deviation method.The connection numbers between the sets were then calculated by using the special value method,subtraction set pair potential method,and semi-partial subtraction set pair potential method.Finally,the similar historical sets were identified according to the same potential principle,and the prediction values were determined by arithmetic average method and weighted average method,and then the annual precipitation prediction model based on set pair analysis was constructed and applied to prediction of the precipitation from 2004 to 2008,and an optimal combination method suitable for constructing the set pair analysis prediction model was proposed.【Results】The results showed that:it was reasonable to determine the historical set by the same potential principle;the accuracy of the prediction value calculated by the arithmetic mean method was acceptable;the subtraction set pair potential and the semi-partial subtraction set pair potential were superior to the special value method in determining the connection numbers;the prediction errors of the three grading-criteria methods were all within 30.00%,and the average relative error of the mean deviation method was 11.58%,the anomaly percentage method was 14.01%,and the mean standard deviation method was 13.74%.【Conclusions】It is feasible to use the set pair analysis method for annual precipitation prediction;the subtraction set pair potential and the semi-partial subtraction set pair potential take into account both the physical characteristics and the dynamic change attributes,and the mean deviation method takes into account the central tendency and the dispersion degree of the data,which is more suitable for the construction of prediction models based on the set pair analysis.
作者
沈瑞
蒋尚明
金菊良
张明
李征
SHEN Rui;JIANG Shangming;JIN Juliang;ZHANG Ming;LI Zheng(Anhui&Huaihe River Institute of Hydraulic Research,Hefei 230088,China;Key Laboratory of Water Conservancy and Water Resources of Anhui Province,Bengbu 233000,China;School of Civil Engineering,Hefei University of Technology,Hefei 230009,China;School of Architecture and Civil Engineering,Anhui Polytechnic University,Wuhu 241000,China)
出处
《华北水利水电大学学报(自然科学版)》
北大核心
2025年第6期56-64,118,共10页
Journal of North China University of Water Resources and Electric Power:Natural Science Edition
基金
国家自然科学基金项目(42271037,52209002)
安徽省自然科学基金项目(2208085US03,2308085US06)
水利部2021年度水利青年拔尖人才(JHQB202227)
安徽省·水利部淮河水利委员会水利科学研究院青年科技创新计划(KY202203)。
关键词
年降水量
相似预测
集对分析
等级标准划分方法
减法集对势
半偏减法集对势
annual precipitation
similarity prediction
set pair analysis
grading-criteria methods
subtraction set pair potential
semi-partial subtraction set pair potential
