针对长江中游城市群洪涝灾害应急转移人口预测中存在的精度瓶颈问题,提出了广义线性模型(generalized linear model,GLM)-空间滞后模型(spatial autoregressive model,SAR)联合预测框架,通过log-link函数与负二项分布解决灾害数据的右...针对长江中游城市群洪涝灾害应急转移人口预测中存在的精度瓶颈问题,提出了广义线性模型(generalized linear model,GLM)-空间滞后模型(spatial autoregressive model,SAR)联合预测框架,通过log-link函数与负二项分布解决灾害数据的右偏分布与过离散问题,构建了以水位超警幅度为核心解释变量的单维度预测模型,并引入空间滞后模型(SAR)量化流域内应急响应的协同效应。基于长江中游三省2019—2023年面板数据的实证结果表明:该模型在跨区域预测中的误差率≤9.2%,较传统线性回归模型降低了8.2个百分点,极端事件的预测误差率压缩至6.7%;揭示了地理异质性对参数的驱动规律,并通过参数迁移模型实现了跨区域适配;通过单维度设计,基层响应效率得到了显著提升,决策链条缩短了4.5 h。研究的创新性体现在“分布修正-空间耦合-动态适配”三重机制,为高波动右偏态灾害数据建模提供了兼具解释力与稳健性的新范式,并为长江中游差异化防灾策略提供了科学支撑。展开更多
We study the quasi likelihood equation in Generalized Linear Models(GLM)with adaptive design∑(i=1)^n xi(yi-h(x'iβ))=0,where yi is a q=vector,and xi is a p×q random matrix.Under some assumptions,it is shown ...We study the quasi likelihood equation in Generalized Linear Models(GLM)with adaptive design∑(i=1)^n xi(yi-h(x'iβ))=0,where yi is a q=vector,and xi is a p×q random matrix.Under some assumptions,it is shown that the Quasi-Likelihood equation for the GLM has a solution which is asymptotic normal.展开更多
目的探讨不同种类毒蛇咬伤的临床特征及预后关键因素,为优化诊疗流程与区域资源配置提供循证依据。方法通过回顾性分析粤港澳蛇伤诊疗中心2023年1月至2025年1月收治的213例毒蛇咬伤患者数据,结合单因素及多因素统计模型,探讨影响预后的...目的探讨不同种类毒蛇咬伤的临床特征及预后关键因素,为优化诊疗流程与区域资源配置提供循证依据。方法通过回顾性分析粤港澳蛇伤诊疗中心2023年1月至2025年1月收治的213例毒蛇咬伤患者数据,结合单因素及多因素统计模型,探讨影响预后的关键因素。结果眼镜蛇咬伤(住院时间中位数10 d vs.竹叶青蛇5 d,P=0.01)、男性患者(白细胞恢复时间7 d vs.女性6 d,P=0.04)及伤口坏死(住院时间中位数8 d,P=0.01)显著延长治疗周期,而早期使用抗蛇毒血清(≤6 h)使凝血功能恢复时间缩短32%(P=0.005,OR=0.028)。多因素广义线性模型(GLM)表明,院前切口引流为极高危因素(OR=387.882,P<0.001),未封闭治疗(OR=1.351,P=0.006)和重度中毒(OR趋近0,P<0.001)显著增加并发症风险。结论眼镜蛇咬伤、男性患者及伤口坏死显著延长毒蛇咬伤患者的治疗周期,而早期使用抗蛇毒血清(≤6 h)可有效缩短凝血功能恢复时间并降低并发症风险。多因素模型进一步揭示,院前切开引流为极高危操作(OR=387.882),未封闭治疗显著延长D二聚体恢复时间和重度中毒显著增加住院时间。临床应严格避免非规范院前处理(如切开引流),优先对高危患者(眼镜蛇咬伤、男性)实施早期血清干预与封闭治疗,并结合蛇类活动规律优化区域救治资源配置,以降低致残率与医疗负担。展开更多
Doline susceptibility mapping(DSM)in karst aquifer is important in terms of estimating the vulnerability of the aquifer to pollutants,estimating the infiltration rate,and infrastructures exposed to the development of ...Doline susceptibility mapping(DSM)in karst aquifer is important in terms of estimating the vulnerability of the aquifer to pollutants,estimating the infiltration rate,and infrastructures exposed to the development of dolines.In this research,doline susceptibility map was prepared in Saldaran mountain by generalized linear model(GLM)using 14 affecting parameters extracted from satellite images,digital elevation model,and geology map.Only 8 parameters have been inputted to the model which had correlation with dolines.In this regards,306 dolines were identified by the photogrammetric Unmanned Aerial Vehicles(UAV)method in 600 hectares of Salderan lands and then,these data were divided into the training(70%)and testing(30%)data for modelling.The results of DSM modeling showed that classified probability of doline occurrences in the Saldaran mountain were as follow:16.5%of the area high to very high,72%in the class of low to very low,and 5%in the moderate class.Also,locally,in Saldaran mountain,the Pirghar aquifer has the highest potential for the doline development,followed by Bagh Rostam and Sarab aquifers.Also,the precipitation,digital elevation model,Topographic Position Index,drainage density,slope,TRASP(transformed the circular aspect to a radiation index),Snow-Covered Days and vegetation cover index are of highest importance in the DSM modeling,respectively.Accurate evaluation of the model using the Receiver Operating Characteristics(ROC)curve represents a very good accuracy(AUC=0.953)of the DSM model.展开更多
The present study investigates the wavespace of Highly Contrasted Structures(HCS)and Highly Dissipative Structures(HDS)by wave-based models.The Asymptotic Homogenization Method(AHM),exploits the asymptotic Zig-Zag mod...The present study investigates the wavespace of Highly Contrasted Structures(HCS)and Highly Dissipative Structures(HDS)by wave-based models.The Asymptotic Homogenization Method(AHM),exploits the asymptotic Zig-Zag model and homogenization technique to compute the bending wavenumbers via a 6th-order equation.The General Laminate Model(GLM)employs Mindlin’s displacement field to establish displacement-constraint relationships and resolves a quadratic Eigenvalue Problem(EVP)of the dispersion relation.The Wave Finite Element(WFE)scheme formulates the Nonlinear Eigenvalue Problem(NEP)for waves in varying directions and tracks complex wavenumbers using Weighted Wave Assurance Criteria(WWAC).Two approaches are introduced to estimate the Damping Loss Factor(DLF)of HDS,with the average DLF calculated by the modal density at various angles where non-homogeneity is present.Evaluation of robustness and accuracy is made by comparing the wavenumbers and DLF obtained from AHM and GLM with WFE.WFE is finally extended to a sandwich metastructure with a non-homogeneous core,and the Power Input Method(PIM)with Finite Element Method(FEM)data is employed to assess the average DLF,demonstrating an enhanced DLF compared to layered configurations with the same material portion,indicating increased energy dissipation due to the bending-shear coupling effects.展开更多
基于Tweedie分布的广义线性模型(generalized linear model,简称GLM),并结合Kriging模型,发展了日降水量统计降尺度的GLM-Kriging模型.首先用GLM拟合研究区域内日降水量与数值模式输出的影响局地降水的物理量之间的关系,日降水量的空间...基于Tweedie分布的广义线性模型(generalized linear model,简称GLM),并结合Kriging模型,发展了日降水量统计降尺度的GLM-Kriging模型.首先用GLM拟合研究区域内日降水量与数值模式输出的影响局地降水的物理量之间的关系,日降水量的空间相关性反映在模型的残差中;然后用Kriging模型来拟合GLM的随机化百分位残差(randomized quantile residuals,简称RQ残差).结合NCEP再分析资料应用于2007年7月沂沭泗流域的42站日降水观测,结果表明GLM-Kriging降尺度模型较好地还原了主要降水过程,整体上取得了较高的准确度,可用于气候变化影响评估或数值天气预报产品的释用,还可进一步扩展为日降水量的时空统计模型.展开更多
基金Supported by the National Natural Science Foundation of China(10371092)
文摘We study the quasi likelihood equation in Generalized Linear Models(GLM)with adaptive design∑(i=1)^n xi(yi-h(x'iβ))=0,where yi is a q=vector,and xi is a p×q random matrix.Under some assumptions,it is shown that the Quasi-Likelihood equation for the GLM has a solution which is asymptotic normal.
文摘目的探讨不同种类毒蛇咬伤的临床特征及预后关键因素,为优化诊疗流程与区域资源配置提供循证依据。方法通过回顾性分析粤港澳蛇伤诊疗中心2023年1月至2025年1月收治的213例毒蛇咬伤患者数据,结合单因素及多因素统计模型,探讨影响预后的关键因素。结果眼镜蛇咬伤(住院时间中位数10 d vs.竹叶青蛇5 d,P=0.01)、男性患者(白细胞恢复时间7 d vs.女性6 d,P=0.04)及伤口坏死(住院时间中位数8 d,P=0.01)显著延长治疗周期,而早期使用抗蛇毒血清(≤6 h)使凝血功能恢复时间缩短32%(P=0.005,OR=0.028)。多因素广义线性模型(GLM)表明,院前切口引流为极高危因素(OR=387.882,P<0.001),未封闭治疗(OR=1.351,P=0.006)和重度中毒(OR趋近0,P<0.001)显著增加并发症风险。结论眼镜蛇咬伤、男性患者及伤口坏死显著延长毒蛇咬伤患者的治疗周期,而早期使用抗蛇毒血清(≤6 h)可有效缩短凝血功能恢复时间并降低并发症风险。多因素模型进一步揭示,院前切开引流为极高危操作(OR=387.882),未封闭治疗显著延长D二聚体恢复时间和重度中毒显著增加住院时间。临床应严格避免非规范院前处理(如切开引流),优先对高危患者(眼镜蛇咬伤、男性)实施早期血清干预与封闭治疗,并结合蛇类活动规律优化区域救治资源配置,以降低致残率与医疗负担。
文摘Doline susceptibility mapping(DSM)in karst aquifer is important in terms of estimating the vulnerability of the aquifer to pollutants,estimating the infiltration rate,and infrastructures exposed to the development of dolines.In this research,doline susceptibility map was prepared in Saldaran mountain by generalized linear model(GLM)using 14 affecting parameters extracted from satellite images,digital elevation model,and geology map.Only 8 parameters have been inputted to the model which had correlation with dolines.In this regards,306 dolines were identified by the photogrammetric Unmanned Aerial Vehicles(UAV)method in 600 hectares of Salderan lands and then,these data were divided into the training(70%)and testing(30%)data for modelling.The results of DSM modeling showed that classified probability of doline occurrences in the Saldaran mountain were as follow:16.5%of the area high to very high,72%in the class of low to very low,and 5%in the moderate class.Also,locally,in Saldaran mountain,the Pirghar aquifer has the highest potential for the doline development,followed by Bagh Rostam and Sarab aquifers.Also,the precipitation,digital elevation model,Topographic Position Index,drainage density,slope,TRASP(transformed the circular aspect to a radiation index),Snow-Covered Days and vegetation cover index are of highest importance in the DSM modeling,respectively.Accurate evaluation of the model using the Receiver Operating Characteristics(ROC)curve represents a very good accuracy(AUC=0.953)of the DSM model.
基金supported by the Natural Sciences and Engineering Research Council of Canada-Discovery Grant(individual)Program(No.NSEC-DG#355433-2009)funded by the LabEx CeLyA(Centre Lyonnais d’Acoustique,No.ANR-10-LABX-0060)of Universite?de Lyon。
文摘The present study investigates the wavespace of Highly Contrasted Structures(HCS)and Highly Dissipative Structures(HDS)by wave-based models.The Asymptotic Homogenization Method(AHM),exploits the asymptotic Zig-Zag model and homogenization technique to compute the bending wavenumbers via a 6th-order equation.The General Laminate Model(GLM)employs Mindlin’s displacement field to establish displacement-constraint relationships and resolves a quadratic Eigenvalue Problem(EVP)of the dispersion relation.The Wave Finite Element(WFE)scheme formulates the Nonlinear Eigenvalue Problem(NEP)for waves in varying directions and tracks complex wavenumbers using Weighted Wave Assurance Criteria(WWAC).Two approaches are introduced to estimate the Damping Loss Factor(DLF)of HDS,with the average DLF calculated by the modal density at various angles where non-homogeneity is present.Evaluation of robustness and accuracy is made by comparing the wavenumbers and DLF obtained from AHM and GLM with WFE.WFE is finally extended to a sandwich metastructure with a non-homogeneous core,and the Power Input Method(PIM)with Finite Element Method(FEM)data is employed to assess the average DLF,demonstrating an enhanced DLF compared to layered configurations with the same material portion,indicating increased energy dissipation due to the bending-shear coupling effects.
文摘基于Tweedie分布的广义线性模型(generalized linear model,简称GLM),并结合Kriging模型,发展了日降水量统计降尺度的GLM-Kriging模型.首先用GLM拟合研究区域内日降水量与数值模式输出的影响局地降水的物理量之间的关系,日降水量的空间相关性反映在模型的残差中;然后用Kriging模型来拟合GLM的随机化百分位残差(randomized quantile residuals,简称RQ残差).结合NCEP再分析资料应用于2007年7月沂沭泗流域的42站日降水观测,结果表明GLM-Kriging降尺度模型较好地还原了主要降水过程,整体上取得了较高的准确度,可用于气候变化影响评估或数值天气预报产品的释用,还可进一步扩展为日降水量的时空统计模型.
