无单元法用于地震波波动方程模拟与成像

波动方程方法是解决地震正反演问题的基本工具之一.无单元法作为一种新兴的偏微分方程数值计算方法,已经在材料力学、热传导等领域取得了显著的成功.由于抛弃了单元的概念及采用滑动最小二乘的拟合方法,使得无单元法具有前处理简单、精度高、独立变量解高次连续等优点.本文首先介绍无单元法求解波动方程的原理,指出影响其精度的主要因素.在算例的基础上详细讨论了无单元法用于实际波动问题的效果,并进一步尝试利用无单元法进行地震波数值模拟和反演成像的研究.模型计算的结果表明,无单元法能够较好的处理地震模拟和成像问题,精度和稳定性是令人满意的.     

复杂卫星图像中的小目标船舶识别

船舶作为海上的重要目标,实现对船舶自动识别有重要的意义。针对卫星图像中云雾、海岸背景等复杂海情对船舶识别带来的干扰,以及小目标船舶高漏检率问题,本文提出一种多尺度深度学习模型训练策略,在此基础上构建了一种船舶识别的深度学习网络,该网络可分为多尺度训练、特征提取、生成目标建议区域、船舶分类这4个部分。首先,采用多尺度的训练策略,将多尺度的船舶样本送入网络中进行训练,这样在训练样本中加入了大量小目标船舶的样本,使网络充分提取到小目标船舶的特征;其次,通过卷积神经网络对目标船舶进行特征自适应提取;然后,目标区域建议网络可依据卷积神经网络提取到的特征,在图像中找到感兴趣目标区域,即框定船舶的位置;最后,通过多个全连接层的组合,将高维特征映射到一个4元组中,再运用分类函数输出每一类船舶的概率值,概率值最大的则为该船舶的类别。同时为解决云雾遮挡和海岸背景的干扰,采用了一种负样本增强学习的方法,在样本数据集中加入了大量只含有云雾和海岸背景的图片,进行负样本扩充,增强网络模型对云雾及海岸背景的特征学习能力,以此解决复杂海情的影响。实验结果表明,所提方法有效解决了复杂海情条件下的船舶识别难,以及小目标船舶识别难的问题,实现了复杂海情条件下的船舶识别。同时,与现有成熟的深度学习目标识别算法相比,本文算法的精确度和召回率分别提升了6.98%和18.17%,所训练的模型具有良好的泛化能力和鲁棒性。     

基于BDS的混合动力船舶节能研究

提出一种基于北斗卫星导航系统(BDS)的油电混合动力推进系统,柴油机和电动机采取PTH驱动方式,采用燃油和电能混合提供推进动力,可以依据BDS导航的航行区域以及航行速度选择不同动力源输出.经过燃油经济性和效益分析,与传统推进系统相比,应用基于BDS的混合动力推进系统,公务船运营成本降低26%,增加的初始建造投资成本3年内可以收回;金枪鱼延绳钓船综合节油15%~20%,通过节油方式可在4年内收回初始建造投资成本.基于BDS的混合动力系统可以增加推进冗余度,提高船舶安全性,具有良好的节能减排和节约运营成本的效果.     

Application of Line Boundary Technique to 2D Tidal Current Simulation

To deal with the problems concerning the shore boundary,moving boundary and engineeringboundary which are encountered frequently in 2D tidal current simulation by the finite difference method,theconcept of line boundary is introduced and studied here,and then the line boundary technique in common useis proposed in this paper.Analysis of some calculation cases shows that this technique is practical,effective,and simple in 2D tidal current simulation involving different boundaries.     

大连港外海域船舶尾迹光学特征研究

通过大连港外海上实验测量船舶含气泡尾迹表观光学特征(如离水辐亮度、遥感反射率),分析了尾迹气泡对极度混浊的Ⅱ类水体海水表观光学特征的影响.实验结果证明:无论在可见光还是近红外波段,尾迹气泡都增强了海水表观光学量,如"耘海"船尾迹遥感反射率在可见光波段相对背景海水提高了50%,在近红外波段相对提高了159%.数据分析结果表明在此种水体中,气泡使得尾迹海水水色变黄,而不是在Ⅰ类水体中使水色变绿.随着观测点在尾迹中的位置靠近目标船,气泡数密度不断增加,遥感反射率也随之增强.     

Numerical study of sand scour with a modified Eulerian model based on incipient motion theory

Application of the standard Eulerian model to simulations of sand scour results in unrealistic phenomena. Therefore, the present work develops a modified Eulerian model based on sand incipient motion theory. The modified model is applied for simulating a two-dimensional single vertical jet and a moving planar jet. The simulation results generally demonstrate fairly good agreement with published results of scour profiles and the velocity contours of the water and sand phases. In addition, equations to describe self-similar scour profiles for the moving planar jet cases are given. The results demonstrate that the modified model efficiently and accurately simulates the two-dimensional sand scour produced by jets, particularly for the moving jet cases.     

基于分区间泛化Kriging近似模型的船舶局部结构形状优化

为提高Kriging近似模型在船舶结构性能多维度响应预测方面的适用性,对常规Kriging近似模型进行分区间泛化改进:一是对设计样本点的各个维度(分量)进行划区,并在每一个分区间内采用最优拉丁超立方(OLhd)取样;二是引入比例系数w1组合高斯与指数型相关函数提高模型对数据的泛化能力,从而在每个划分的区间内建立泛化的Kriging近似模型。通过SCH测试函数,验证了构造的分区间泛化Kriging模型有效性。结合参数化建模和改进Kriging近似模型对某过渡肘板和舱口角隅边界进行形状优化,优化第一阶段由改进的近似模型通过多岛遗传算法得到全局初步的最优解,第二阶段在初步解的基础上缩小优化变量范围,由少量的FEM计算即可在小范围内搜寻到精确的最优设计变量。结果表明:分区间泛化Kriging近似模型在预测多维度响应时较常规Kriging模型预测精度更高;分阶段的形状优化流程在保证极小误差和缩小计算成本的情况下能够得到理想的应力分布和重量优化结果,有助于船舶结构的轻量化研究。     

Population abundance and seasonal migration patterns indicated by commercial catch-per-unit-effort of hakes (Merluccius capensis and M. paradoxus) in the northern Benguela Current Large Marine Ecosystem

We developed generalised additive models (GAMs) to estimate standardised time-series of population abundance indices for assessment purposes and to infer ecological and behavioural information on northern Benguela hakes, Merluccius capensis and M. paradoxus, using haul-by-haul commercial trawl catch-rate data as proxies for hake densities. The modelling indicated that individual ship identifiers should be used rather than general vessel characteristics, such as vessel size. The final models explained 79% and 68% of the variability in the commercial catch rates of M. capensis and M. paradoxus, respectively. The spatial density patterns were consistent and confirmed existing knowledge about these species in the northern Benguela system. Furthermore, seasonal migration patterns were described for the first time and were found to correspond to the known spawning areas and seasons for M. capensis and M. paradoxus. Spatial density patterns were validated using the geostatistical modelling results of fisheries-independent trawl survey data. Improved understanding of the relationships between fleet dynamics and fish movement can be achieved by taking into consideration the present catch-rate model and spatial and seasonal distribution maps. We conclude that the yearly standardised CPUE time-series are problematic as proxies for total stock abundance because of spatial coverage issues. Consequently, such CPUE data should not be used for stock-size assessments and fisheries advice concerning northern Benguela hakes until this is solved. We generally recommend the exclusion of standardised CPUE time-series from stock assessments when important and changing parts of the stock distribution cannot be targeted by the fishery, such as due to closed areas or seasons.     

基于结构特征的SAR船只类型识别能力分析

1 引言 SAR是船只探测的有效手段,利用SAR影像不仅可以检测船只目标,而且还可以提取船只的长度、宽度以及航速、航向等信息.随着星载SAR空间分辨率的提高,使基于星载SAR影像的船只类型识别成为可能.利用星载SAR可以检测出大面积范围内的船只和初步识别其类型.     

�����������񻯷����Ƚ�

???Shepard???????????????????????????????Kriging????????????????п???????????????????????????????????????ε?????????????????????÷?Χ??????3????????б???????????1??????????????????????????????????????????????????2?????Kriging?????????????????????????????????????????仯???????????????????????????????????????仯С??????