基于投影寻踪的洪水灾情评价插值模型

Interpolation Model for Evaluating Loss of Flood Disaster Based on Genetic Projection Pursuit

为解决单项洪水灾情等级评价指标评价结果的不相容问题,提高综合评价各层次的分辩力和评价模型的精度,本文利用投影寻踪、遗传算法、插值型曲线,为洪水灾情评价提供了一种新模型——遗传投影寻踪插值模型。通过最佳投影向量和评价指标向量的内积可把洪水灾情多维评价样本指标综合成一维投影指标,根据该投影指标值和对应等级的分布,可建立洪水灾情评价的插值模型,解决了各单项洪水灾情评价指标评价结果的不相容问题,提高了洪水灾情综合评价问题各层次的分辩力。作为示例,对中国部分省市的洪水灾情进行了综合评价。

In order to solve the incompatible indexes problem and to raise the model precision of loss of flood disaster level assessment, a new model, genetic projection pursuit interpolation model (GPPIM), is presented for loss of flood disaster level assessment with lots of data, projection pursuit, genetic algorithm, interpolation shape curve and the assessment standards of loss of flood disaster level. The multidimensional indexes of loss of flood disaster can be synthesized with one dimension projection index by the inter product between the optimum projection vector and assessment index vector. The interpolation model for assessing loss of flood disaster is established by the distributions between index value and corresponding of grades. It solves the problem of incompatible assessment indexes. And its detailed step is developed. As an example, loss of flood disaster level is assessed by GPPIM in China.