【摘要】 科学选取灾情评价指标,构造基于组合权重的洪涝灾情指数序列,利用样本的模糊信息,通过集值化的模糊数学处理,建立灾情指数的超概率评估模型。同时,采用粒子群算法以及两阶段寻优方法求解模型中的参数r,以构造基于粒子群的灾情指数与灾变年的非线性伯努利预测模型,并将预测模型运用于广西洪涝灾情指数和灾变年预测中.结果表明,广西发生中灾以上洪涝灾害年份的超越概率是0.3175.从2013年起五年内,发生中灾以上年份是2014年和2017年,与实际灾情拟合较好;此外,将构造的伯努利模型与GM(1,1)模型和Verhulst模型进行对比分析后发现,非线性伯努利模型更适合做洪涝灾情指数预测和灾变年预测.更多还原

【Abstract】 The flood disaster evaluation index is selected in scientific way.The indicator for establishing index series of flood disaster by using combination weight.An exceeded probability model of flood disaster index is developed with the fuzzy information of samples which is processed with set-valued fuzzy mathematics.At the same time,a nonlinear grey bernoulli model of flood disaster index and hazard-year based on particle swarm optimization are constructed with the parameter r that is calculated by PSO and two-stage optimization method,and then applied to the prediction of Guangxi’s flood disaster index and hazard-year.Results show the excursion probability of exceeding a moderate flood hazard year in Guangxi is 0.3175.Within five years from 2013,the years whose flood disaster is more serious than middle disaster year’s is 2014 and 2017,which is in a good coincidence with the actual.In addition,Comparing with the GM(1,1) model and Verhulst model,nonlinear grey bernoulli model is more suitable for flood disaster index in forecasting and hazard-year prediction.更多还原