摘要
针对投影寻踪(PP)模型在实际应用中最佳投影方向α难以确定的不足,利用蝙蝠算法(BA)搜寻PP模型最佳投影方向α,提出BA-PP评估模型,并同PSO-PP模型作对比,以文山州1990—2013年旱灾评估为例进行实例研究,选取农作物受灾面积、成灾面积、粮食损失和经济作物损失4个旱灾评估基本指标,并利用旱灾投影系列均值Z及标准差σ构造旱灾评估分级标准对实例进行评估。结果表明:1BA算法收敛精度优于PSO算法,具有较高的寻优精度和全局寻优能力,将BA算法用于PP模型佳投影方向α的选取,可有效提高PP模型评估精度,避免最佳投影方向α寻优结果变化范围过大的缺陷;2BA-PP模型对文山州24 a来旱灾评估结果为特大旱灾3次,重大旱灾5次,较大旱灾16次,平均3 a发生一次重大旱灾,旱灾程度随时间呈增加趋势,且增加趋势显著。
According to the projection pursuit( PP) deficiency model in practical application due to the best projection direction, using the bat algorithm( BA) to search for the best projection direction PP model, proposes the BA-PP evaluation model, and compared with the PSO-PP model, by the state of Wenshan in 1990—2013 drought evaluation as an example for case study, selects the crops disaster area, the disaster area, the loss of grain and economic crop loss 4 drought assessment basic index, and differential tectonic drought assessment grading standards for example is evaluated by a drought projection series of mean and standard. The results showed that: 1. The convergence of the BA algorithm has higher precision than the PSO algorithm, the optimization accuracy and the ability of global optimization, BA algorithm is used to select the PP model is the best projection direction, can effectively improve the accuracy of PP evaluation model, avoid the best projection direction optimization results, the defect of large variation range. The BA-PP model of Wenshan 24 years of drought drought assessment results for 3 times, 5 times larger for major drought, drought 16 times, an average of3 years and a major drought, drought degree increased with the increase of time, and a significant increase trend.
出处
《人民珠江》
2016年第9期26-30,共5页
Pearl River
关键词
旱灾评估
投影寻踪
蝙蝠算法
文山州
drought assessment
projection pursuit
bat algorithm
Wenshan State
