基于集对分析的水资源系统预测方法及其应用

发布时间：2018-04-24 19:40

本文选题：水资源系统 + 预测　；参考：《合肥工业大学》2017年硕士论文

【摘要】：水是人类赖以生存和发展的不可替代的重要资源。进入21世纪以来,许多国家都面临着水资源危机的挑战,水资源短缺、水体污染、洪涝灾害等严重阻碍了我国经济和社会的发展,造成一系列社会问题。研究水资源预测问题,对实现水资源的可持续利用、促进水资源与经济、社会和生态环境的可持续发展、缓解我国水资源危机均具有十分重要的意义。在水资源预测过程中的主要问题是如何有效处理不确定性,而集对分析方法在处理这些问题时表现出明显的优势。基于此,论文主要在集对分析理论的基础上,研究了水资源系统预测方法及其应用问题,取得如下研究成果:(1)针对目前研究中径流级别划分存在不确定性,主张在径流分级时应该考虑选取合适的径流预测模型且对径流资料进行三性审查,以确保预测的准确程度。选定历史年径流资料的频率分布曲线,根据年径流的累计频率在频率曲线上确定丰、平、枯状态划分的临界值,从而确定各年均径流量的状态。建立历史径流集对和当前径流集对,并与BP神经网络方法进行耦合从而实现对径流的预测,结果表明取得的预测结果良好。(2)考虑到集对分析方法在径流等级划分上存在主观性,提出模糊化径流分级标准,并引入年内径流分配的丰枯贡献权重因子,计算出年径流对于各个等级的综合隶属度,并建立综合隶属度和理想分级隶属度的集合,引进模糊数学中贴近度的概念,运用模糊识别的方法对径流集合中的各元素进行分类,通过比较贴近度的大小进行最终判断。将该分类方法应用于年径流预测中,结果表明所得的预测结果相对误差较小,该分类方法准确、有效。(3)为分析识别影响城市用水量的相关影响因子,运用灰色关联分析方法计算各影响因子的灰色关联度,通过比较灰色关联度的大小确定对城市用水量影响较大的影响因子,并利用灰色关联度来计算各个影响因子所占的权重。在建立集对分析聚类预测模型中,分别通过灰色关联度和层次分析法这两种方法计算每个影响因素的权重,建立待预测样本与参照系统之间的联系度,进而预测出待预测年份的用水量。经计算前者具有较高的精度,是预测城市用水量较理想的方法。
[Abstract]:Water is an irreplaceable important resource for human survival and development. Since the beginning of the 21st century, many countries are facing the challenge of water resources crisis. Water shortage, water pollution and flood disasters have seriously hindered the economic and social development of our country and caused a series of social problems. It is of great significance to study the prediction of water resources to realize the sustainable utilization of water resources, to promote the sustainable development of water resources and economic, social and ecological environment, and to alleviate the crisis of water resources in China. The main problem in the process of water resources prediction is how to deal with uncertainty effectively, and the set pair analysis method has obvious advantages in dealing with these problems. Based on this, based on the theory of set pair analysis, the prediction method of water resources system and its application are studied in this paper. The following research results are obtained: (1) there is uncertainty in the classification of runoff levels in the present research. In order to ensure the accuracy of runoff prediction, it is suggested that the appropriate runoff forecasting model should be selected and the runoff data should be examined three ways in order to ensure the accuracy of the prediction. The frequency distribution curve of historical annual runoff data is selected and the critical value of annual runoff annual runoff is determined on the frequency curve according to the cumulative frequency of annual runoff. The historical runoff set pair and the current runoff set pair are established and coupled with BP neural network method to realize runoff prediction. The results show that the predicted results are good. The fuzzy runoff classification standard is put forward, and the weight factor of the annual runoff distribution is introduced to calculate the comprehensive membership degree of the annual runoff to each grade, and to establish the set of the comprehensive membership degree and the ideal grade membership degree. This paper introduces the concept of closeness in fuzzy mathematics, classifies the elements in runoff set by fuzzy recognition method, and makes the final judgment by comparing the size of closeness. The results show that the relative error of the predicted results is small, and the classification method is accurate and effective. It is an effective factor to analyze and identify the influence factors of urban water consumption. The grey correlation degree of each influence factor is calculated by using the method of grey correlation analysis, and the influence factor on urban water consumption is determined by comparing the size of grey correlation degree, and the weight of each influence factor is calculated by using grey correlation degree. In the cluster prediction model of set pair analysis, the weight of each influencing factor is calculated by grey correlation degree and analytic hierarchy process, and the relation between the sample to be predicted and the reference system is established. Furthermore, the water consumption of the year to be forecasted is predicted. The former has higher accuracy and is an ideal method for predicting urban water consumption.
【学位授予单位】：合肥工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TV213.4

【参考文献】