This paper focuses on the simulation analysis of stripe formation and dynamic features of intersecting pedestrian flows.The intersecting flows consist of two streams of pedestrians and each pedestrian stream has a des...This paper focuses on the simulation analysis of stripe formation and dynamic features of intersecting pedestrian flows.The intersecting flows consist of two streams of pedestrians and each pedestrian stream has a desired walking direction.The model adopted in the simulations is the social force model, which can reproduce the self-organization phenomena successfully. Three scenarios of different cross angles are established. The simulations confirm the empirical observations that there is a stripe formation when two streams of pedestrians intersect and the direction of the stripes is perpendicular to the sum of the directional vectors of the two streams. It can be concluded from the numerical simulation results that smaller cross angle results in higher mean speed and lower level of speed fluctuation. Moreover, the detailed pictures of pedestrians＇ moving behavior at intersections are given as well.展开更多
【目的】筛选整粒小麦籽粒蛋白质的近红外特征光谱波段并建立优化模型,可实现快速、无损测定整粒小麦籽粒蛋白质含量,为田间便携式小麦籽粒蛋白质含量速测仪设计提供依据。【方法】2012—2013年以蛋白质含量有明显差异的8个冬小麦品种...【目的】筛选整粒小麦籽粒蛋白质的近红外特征光谱波段并建立优化模型,可实现快速、无损测定整粒小麦籽粒蛋白质含量,为田间便携式小麦籽粒蛋白质含量速测仪设计提供依据。【方法】2012—2013年以蛋白质含量有明显差异的8个冬小麦品种为试验品种,设置3个施氮量和2个灌溉量共6个处理,建立丰富的样本类型,共采集176个小麦籽粒光谱数据;将ASD Field Spec Pro光谱仪采集到的基于全反射下垫面的整粒小麦籽粒反射光谱通过公式A=log(1/R)转换为吸收光谱,对吸收光谱采用S-G平滑、多元散射校正和基线校正等方法进行预处理,以消除背景噪声,然后采用交叉验证偏最小二乘回归方法进行特征波段压缩;分析比较无信息变量剔除法(UVE)结合交叉验证偏最小二乘回归、连续投影算法(SPA)结合交叉验证偏最小二乘回归、UVE与SPA组合后结合交叉验证偏最小二乘回归、UVE与SPA组合后结合多元线性回归(MLR)及UVE与SPA组合后结合逐步多元线性回归(SMLR)等多种特征光谱筛选方法选出的蛋白质特征波段的优劣,并与凯氏定氮法测定的小麦籽粒蛋白质含量进行回归分析,构建并优选小麦籽粒蛋白质最佳预测模型。【结果】利用无信息变量剔除(UVE)方法可将与小麦籽粒蛋白质含量无关的信息变量剔除,把籽粒的原始光谱由1 621个波段压缩至717个,在保留了蛋白质信息的同时,实现了特征谱段的初次优选;对逐步多元线性回归(SMLR)、连续投影算法(SPA)、连续投影算法(SPA)+逐步多元线性回归(SMLR)及连续投影算法(SPA)+偏最小二乘回归(PLS)+交叉验证(CV)等特征波段优选算法比较发现,不同的方法获得的特征谱段有差异,构建的模型及精度也明显不同。对经过无信息变量剔除(UVE)法筛选光谱特征谱段,利用SPA消除光谱矩阵中波段共线性影响,再利用SMLR筛选出小麦籽粒蛋白质信息贡献最大的15个特�展开更多
基金Project supported by the National Natural Science Foundation of China(Grant No.61233001)the Fundamental Research Funds for the Central Universities,China(Grant No.2017JBM014)
文摘This paper focuses on the simulation analysis of stripe formation and dynamic features of intersecting pedestrian flows.The intersecting flows consist of two streams of pedestrians and each pedestrian stream has a desired walking direction.The model adopted in the simulations is the social force model, which can reproduce the self-organization phenomena successfully. Three scenarios of different cross angles are established. The simulations confirm the empirical observations that there is a stripe formation when two streams of pedestrians intersect and the direction of the stripes is perpendicular to the sum of the directional vectors of the two streams. It can be concluded from the numerical simulation results that smaller cross angle results in higher mean speed and lower level of speed fluctuation. Moreover, the detailed pictures of pedestrians＇ moving behavior at intersections are given as well.
文摘【目的】筛选整粒小麦籽粒蛋白质的近红外特征光谱波段并建立优化模型,可实现快速、无损测定整粒小麦籽粒蛋白质含量,为田间便携式小麦籽粒蛋白质含量速测仪设计提供依据。【方法】2012—2013年以蛋白质含量有明显差异的8个冬小麦品种为试验品种,设置3个施氮量和2个灌溉量共6个处理,建立丰富的样本类型,共采集176个小麦籽粒光谱数据;将ASD Field Spec Pro光谱仪采集到的基于全反射下垫面的整粒小麦籽粒反射光谱通过公式A=log(1/R)转换为吸收光谱,对吸收光谱采用S-G平滑、多元散射校正和基线校正等方法进行预处理,以消除背景噪声,然后采用交叉验证偏最小二乘回归方法进行特征波段压缩;分析比较无信息变量剔除法(UVE)结合交叉验证偏最小二乘回归、连续投影算法(SPA)结合交叉验证偏最小二乘回归、UVE与SPA组合后结合交叉验证偏最小二乘回归、UVE与SPA组合后结合多元线性回归(MLR)及UVE与SPA组合后结合逐步多元线性回归(SMLR)等多种特征光谱筛选方法选出的蛋白质特征波段的优劣,并与凯氏定氮法测定的小麦籽粒蛋白质含量进行回归分析,构建并优选小麦籽粒蛋白质最佳预测模型。【结果】利用无信息变量剔除(UVE)方法可将与小麦籽粒蛋白质含量无关的信息变量剔除,把籽粒的原始光谱由1 621个波段压缩至717个,在保留了蛋白质信息的同时,实现了特征谱段的初次优选;对逐步多元线性回归(SMLR)、连续投影算法(SPA)、连续投影算法(SPA)+逐步多元线性回归(SMLR)及连续投影算法(SPA)+偏最小二乘回归(PLS)+交叉验证(CV)等特征波段优选算法比较发现,不同的方法获得的特征谱段有差异,构建的模型及精度也明显不同。对经过无信息变量剔除(UVE)法筛选光谱特征谱段,利用SPA消除光谱矩阵中波段共线性影响,再利用SMLR筛选出小麦籽粒蛋白质信息贡献最大的15个特�
