半参数p-范极大似然回归

应用核权函数,在观测为误差单峰、对称的情况下,得到了一元p-范分布的半参数模型的计算公式.详细推导了p已知时一元p-范分布极大似然方程的解算公式,将半参数回归模型应用到极大似然平差的参数估计理论中,得到了一个比较好的算法.最后,构造了两个模拟平差问题,说明了此方法的优越性.     

基于灰色系统的原煤回采成本预测分析

运用灰色预测理论结合VB编程,对煤炭回采成本中的各组成部分分别建立了预测模型,对平煤集团某煤矿的原煤回采成本进行了预测.经检验并与多项式回归方法及实际回采成本作对比,结果较为理想,使煤炭回采成本的预测更加符合实际,为煤炭回采成本的控制提供可靠的依据.     

基于通量均衡基准的地壳垂直形变场逼近的Fourier分析法

本文根据地壳运动的全球通量均衡假说建立了地壳垂直形变场的分析模型,分析了以往作者所研究的球谐函数和球冠谐函数模型的优缺点,着重推导了在密切平面坐标系下该模型的Fourier级数解,该模型一方面可以简单地应用于地壳垂直形变场的逼近,另一方面为采用谱分析法提取地壳垂直形变场信息提供了理论依据。     

基于HJ-1A高光谱影像的盐渍化土壤信息提取——以渭干河-库车河绿洲为例

以环境小卫星高光谱影像为主要数据源,在野外实测样本的支持下进行光谱反射率及其变换形式与土壤含盐量的相关性分析,筛选盐渍化土壤响应敏感波段,利用曲线回归分析方法,建立基于高光谱影像的新疆渭干河-库车河绿洲土壤含盐量定量反演模型.结果表明:研究区土壤含盐量的影像响应波段基本位于近红外波段,其中以780～924 nm波长范围最佳,相关系数R≈0.8;反射率对数的倒数一阶微分土壤含盐量预测模型精度最高,回归方程为Y=-4.152-27.735X+769.813X2,模型及其检验的决定系数都在0.88以上,均方根误差约为3.该模型的建立可为区域盐渍化土壤信息的提取及监测提供参考.     

基于空间回归分析的城市土地综合承载力研究——以环渤海地区城市群为例

归纳出了影响城市土地综合承载力的3个主要因素;并对普通最小二乘法(OLS)模型和地理加权回归(GWR)模型进行了分析;进而对环渤海地区城市群土地综合承载力进行了实证研究.结果表明:①环渤海地区城市群土地人口承载力的增强,土地资源、环境承载力的优化,土地经济、社会承载力的拉动,能够提升城市土地综合承载力水平.②人口因素、资源环境因素对土地综合承载力影响的平均空间回归系数均从山东半岛城市群开始沿顺时针方向递增,经济社会因素对土地综合承载力影响的平均空间回归系数则从京津冀城市群开始沿逆时针方向递减.③这3个主要因素对土地综合承载力的影响敏感程度都存在较为明显的“高低区域地带”.     

基于HASM-SSOR模型的近60年来中国气温与降水变化趋势模拟

为了改善HASM模型计算速度慢及存储量大等问题,提出了一种快速有效算法(HASM-SSOR),并以高斯合成曲面为例验证了该方法的有效性。利用全国752站1951—2010年的气象观测资料,对近60年的平均气温、降水分别采用多元回归与HASM-SSOR相结合的方法及多项式回归与HASM-SSOR相结合的方法进行模拟与分析;近60年平均气温、降水拟合的复相关系数分别为0.9923、0.9987;并对1951-2010年各年代际的平均气温、降水的变化趋势进行拟合与分析,发现各年代际气温变化并不一致,全国平均气温呈上升趋势,平均上升率为0.35℃/10a,全国降水平均变化率为9.4mm/10a,无明显的变化趋势。     

江苏县域经济要素关系及演化过程的分位数回归

针对经典回归模型的“两难问题”,引入非参数的分位数回归构建了要素关系模型,并以江苏县域经济为案例分析.结果表明:1)相比OLS回归,分位数回归对县域经济总体模拟效果和刻画演化特征能力更优;2)根据变量关系结构,将县域经济发展驱动机制划分为产业结构占优型、一般均衡驱动型和高效均衡驱动型;3)苏锡常各县市受因子波动影响而使得驱动机制发生改变导致演化过程的分位点跃迁,表现出发展路径选择具有多样化的特点.     

An explanation of the large PGA value of 2013 Ms7.0 Lushan earthquake at 51BXD station through topographic analysis

In Ms7.0 Lushan earthquake, a large amount of strong ground motion recordings were collected. In this paper, we analyze the recordings carefully. The abnor- mality of ground motion recordings is identified through a log linear regression. In the station of 51BXD, the PGA value has exceeded 1 g, which is the biggest peak ground acceleration （PGA） value obtained from all recordings in this earthquake. The log linear relation shows the PGA value in this station is abnormally large. As this station is located on the footage of a hill, the topographic amplifi- cation factor is explored in order to explain this abnor- mality. Through 3D numerical modeling using spectral element method with transmitting boundary conditions, the amplification factor is quantized. In this station, the topo- graphic amplification is highly polarized in the direction of East-West which agrees with the empirical recordings. This research result suggests us in future directionality of topographic amplification should be considered in the aseismic design.     

Using live vegetation volume to analyze the effects of plot Pinus massoniana Lamb on water and soil conservation under natural rainfall events

The 3-D spatial distributions of vegetation are of great significance for water and soil conservation but are rarely concerned in literatures. The live vegetation volume （LVV） was used to relate to water/soil loss under 144 natural erosive rainfall events from 2007 to 2010 in a typical water-eroded area of southern China. Quadratic polynomial regression models were established for five pure tree （Pinus massoniana Lamb） plots between LVV and water （rtmoff）/soil conservation effects （RE/SE）. RE/SE corresponds to the ratios of runoff depth/soil loss of the pure tree plots to that of the control plot under each rainfall event. Increasing LVV exhibits descending （DS）, descending-ascending （DA）, ascending-descending （AD）, and ascending （AS） trends in the LVV-RE and LVV-SE curves. The effects of soil conservation on the plots were generally more noticeable than the effects of water conservation, and most of the RE and SE values reflected the positive effects of water and soil conservation. The effects were mainly positive under heavy rains （e.g., rainfall erosivity, R = 140 MJ mm ha-l h, maximum 30 min intensity, I30 = 16 mm h-l）, whereas the effects were mainly negative under light rains （e.g., R = 45 MJ mm ha-1 h, I30 = 8 mm h-l）. The trees＇ water/soil conservation effects notably transformed when rainfall erosivity and intensity were lower than the positive or negative effects to a certain threshold. About 50% rainfall events led to obvious transform effects when LVVs were near 0.5 or 0.6. These results are able to aid in the decision making on the forest reconstruction in water-eroded areas.     

利用支持向量机和高斯过程回归测定水库诱发的地震(英文)

水库诱发地震震级(M)的预测是在地震工程中的一项重要任务。本文采用支持向量机(SVM)和高斯过程回归(GPR)模型根据水库的参数预测了水库诱发地震震级(M)。综合参数(E)和最大的水库深度(H)作为支持向量机和高斯过程回归模型的输入参数。我们给出一个方程确定水库诱发地震震级(M)。将本文开发的支持向量机和建立的高斯过程回归方法与人工神经网络(ANN)方法相比。结果表明,本文研发的支持向量机和高斯过程回归方法是预测水库诱发地震震级(M)的有效工具。