【摘要】 洪水灾情评估是多因素评价问题,它是包括自然、环境、社会经济等多种因素综合作用的结果,必须对各因素进行综合评估,而不能仅依据某单项指标得出评价结论。目前国内外还没有统一的洪水灾情评估指标体系和各指标定量化等级标准。根据投影寻踪技术提出了洪水灾情投影寻踪评估模型,将评价样本的各因素进行线性投影,以最优投影方向所对应的投影特征量作为评价依据,该模型计算过程简单、结果直观,更重要的是评估结果无人为任意性。将这一模型用于1998年洪水的分类和我国部分省市的洪水灾情评估中。实例分析表明,该模型能够很好地对各种洪水灾情进行有效的评估,是一种处理多因素复杂评价问题的新途径。更多还原

【Abstract】 The difficulty frequently encountered in flood disaster evaluation is that there are so many factors included and the complex interrelationship existed among them.The evaluation cannot be achieved according to only one factor,because all the effect factors associated with flood disaster must be thought over.However,up till now,there is no a unified standard of assessment index system in the world.The research shows that projection pursuit model could form a suitable index for overcoming the difficulties in assessment.Projection pursuit is a statistical technique,which is by projecting high-dimensional data onto low-dimensional space and most of data’s feature can still be retained by a main characteristic component.It makes the assessment of high dimensional feature easier.For the first time,an efficient approach with genetic algorithm to find interesting projection is offered in this paper.Using this new model,we succeeded in analyzing1998flood characters in China,and evaluating flood disasters from1989to1990in33provinces in China.更多还原