基于遗传算法的理想区间法在洪水灾情评价中的应用

洪水灾情评价的实质就是建立各洪水灾情评价指标与洪水灾情等级之间的非线性关系,目前在实际评价过程中反映这种关系的信息来源只有洪水灾情评价标准,而后者一般是以区间形式表示的。基于此,提出了基于加速遗传算法的改进理想区间法(AGAIIM)。AGAIIM直接由洪水灾情评价标准样本数据驱动,把利用全部隶属度值信息进行计算的洪水灾情相对等级值作为洪水灾情的评价结果,可避免应用最大隶属度原则进行判断所可能造成的失真,提高了洪水灾情评价的精度。AGAIIM方法直观、简便、通用,可在具有评价标准或具有已知评价指标值及其等级值样本系列的系统综合评价中推广应用。

Application of Genetic Algorithm Based on Ideal Interval Method to Evaluation of Flood Disaster Loss

Flood disaster loss evaluation is to evaluate the damage degree caused by flood disaster according to flood disaster loss evaluation criterions, existing flood disaster loss evaluation index values and disaster loss evaluation model. The result of disaster loss evaluation, named as disaster loss grade, is of important instructional significance to the flood disaster management. Flood disaster loss involves many factors including the natural environment and social economy etc. There are still no uniform evaluation index systems and grade criterions of flood disaster loss internationally. So evaluation problem of flood disaster loss is still one of difficulties and hotspots of researches on flood disaster. The essential of evaluation of flood disaster loss is to establish the nonlinear relation between flood disaster loss indexes and flood disaster loss grades. Nowadays only evaluation criterion of flood disaster loss, which is expressed by interval concept, reflects this relation information in practical evaluation process. On this basis, improved ideal interval method based on the accelerating genetic algorithm developed by the authors, named AGAIIM for short, is presented in this paper for evaluation of flood disaster loss. The modeling of AGAIIM is the key in this paper, which includes four steps as follows. Step 1 is to create sample series of evaluation criterion of flood disaster loss randomly and to standardize the series named as {x(k,j)|k=1,...,nk,j=1,...,nj}, where nk and nj are the number of the samples and evaluation indexes respectively. Step 2 is to compute the distance D(k,i) between criterion sample x(k,j) and ideal interval of criterion grade, where i is criterion grade value of the i th flood disaster loss grade obtained from evaluation criterion table. Step 3 is to compute relative membership degree values r(k,i) of the sample {x(k,j)} relative to criterion grade ideal interval. Step 4 is to comprehensively evaluate flood disaster loss grade. The computation results of the case study can include four terms as follows. (1) As a new method for flood disaster loss evaluation based on improved ideal interval method and the accelerating genetic algorithm, AGAIIM can describe the nonlinear relation between evaluation indexes and flood disaster loss grades very well. (2) AGAIIM belongs to non-function model evaluation methods, its evaluation process is directly driven by the example series produced from evaluation criterion of flood disaster loss, and flood disaster loss relative grades computed by using all membership degrees information are regarded as evaluation results of flood disaster loss, which can avoid distorting when using the principle of maximum membership degree and can heighten evaluation precision of flood disaster loss. (3)The results of AGAIIM are of real values, so its precision is high. (4)AGAIIM is visual, handy and universal, which can widely be applied to system comprehensive evaluation when evaluation criterion or samples of evaluation indexes and evaluation result values are known.