Seasonal variability of the Black Sea chlorophyll-a concentration

The optimal spectral decomposition (OSD) method is used to reconstruct seasonal variability of the Black Sea horizontally averaged chlorophyll-a concentration from data collected during the NATO SfP-971818 Black Sea Project in 1980–1995. During the reconstruction, quality control is conducted to reduce errors caused by measurement accuracy, sampling strategy, and irregular data distribution in space and time. A bi-modal structure with winter/spring (February–March) and fall (September–October) blooms is uniquely detected and accurately documented. The chlorophyll-a enriched zone rises to 15 m depth in winter and June, and deepens to 40 m in April and 35 m in August. The June rise of the chlorophyll-a enriched zone is accompanying by near-continuous reduction of upper layer maximum chlorophyll-a concentration.     

多弹性体接触问题的数值算法

提出一种新的求解带摩擦的接触问题的数值算法,即拟高斯迭代法。它对法向接触力和切向接触力进行交替迭代,并利用高斯迭代法求解法向接触力,利用分块高斯迭代法求解切向接触力。同其它的数值算法相比,该算法保留了关于接触力的柔度矩阵的稀疏性和对称性,利用矩阵乘向量可以分步进行的技巧,该算法只需存储关于接触力的柔度矩阵的下三角形矩阵的非零元和对角矩阵。根据可能接触边界的分布特点,将区域分解成不同的子区域,引进拉格朗日乘子表示接触力,保证了各子区域的网格剖分和位移求解是完全独立的。基于上述算法和有限元程序自动生成系统开发了相应的求解带摩擦接触问题的软件,数值实验表明,程序是正确的,算法是高效的。     

长方体布局问题的一种启发式算法

长方体布局问题属于NP完全问题,在串行机上解决这类问题只能依赖启发式算法。本文提出了一种对布局空间进行动态分解和对剩余空间进行合并和再利用的启发式算法。分解过程采用三叉树数据结构表示,深度优先原则搜索,根据一定的定序规则和定位规则快速求得问题的满意解。同时针对影响布局结果的"难布置的布局物体",设计剩余空间的合并规则,达到该类物体的布入,并通过算例说明了本算法的有效性,对长方体布局具有一定的指导意义,也为人机结合的优化布局提供了良好基础。     

弱GNSS信号下基于EMD和LSTM的车辆位置预测方法研究

针对弱GNSS环境下组合导航（INS/GNSS）系统存在的定位偏差问题，提出一种基于经验模态分解和长短期记忆网络的车辆位置预测算法。首先，针对训练数据中噪声较大的惯导数据，提出一种融合经验模态分解与离散小波变换的降噪算法。该算法基于噪声能量估计和各阶本征模态函数的功率谱密度函数，提出一种确定混合模态函数阶数上下界的方法，并采用离散小波变换硬阈值法对混合模态函数进行滤波处理，最终利用经过处理的各阶模态函数重构原始数据以达到降噪目的。训练数据经过预处理后，采用改进的堆叠式长短期记忆网络离线训练位置预测模型，利用该训练模型可在线实时进行位置预测。针对车辆定位序贯数据预测，提出一种局部数据降噪方法，该方法利用一定长度时间窗口的历史数据，通过线性最小二乘给出当下时刻数据的预估值，并与实际量测值进行滑动平均滤波，优化位置预测的结果。在封闭场地模拟隧道环境下，对长短期记忆网络输入端进行局部数据降噪与不进行降噪处理比较，经度和纬度的归一化均方误差分别下降了13.34%和9.38%，经度和纬度的归一化平均绝对误差分别下降了8.64%和5.41%；在复杂城市交通环境下，检验提出的方法，经度和纬度的归一化均方误差分别下降了6.51%和5.66%，经度和纬度的归一化平均绝对误差分别下降了5.70%和8.23%。试验结果表明，在弱GNSS信号环境下，提出的车辆位置预测方法有效提高了车辆定位精度和稳定性。     

Airport taxi planning: Lagrangian decomposition

The airport taxi planning (TP) module is a decision tool intended to guide airport surface management operations. TP is defined by a flow network optimization model that represents flight ground movements and improves aircraft taxiing routes and schedules during periods of aircraft congestion. TP is not intended to operate as a stand‐alone tool for airport operations management: on the contrary, it must be used in conjunction with existing departing and arriving traffic tools and overseen by the taxi planner of the airport, also known as the aircraft ground controller. TP must be flexible in order to accommodate changing inputs while maintaining consistent routes and schedules already delivered from past executions. Within this dynamic environment, the execution time of TP may not exceed a few minutes. Classic methods for solving binary multi‐commodity flow networks with side constraints are not efficient enough; therefore, a Lagrangian decomposition methodology has been adapted to solve it. We demonstrate TP Lagrangian decomposition using actual data from the Madrid‐Barajas Airport. Copyright © 2011 John Wiley & Sons, Ltd.     

多学科优化设计在半潜式钻井平台设计中的应用

多学科优化设计（Multidisciplinary Design Optimization，MDO）是通过考虑学科之间的相互耦合来挖掘设计潜力，通过系统的综合分析来进行方案的选择和评估的一种应用于工程设计阶段的先进设计理念和设计方法，能够达到提高设计效率，降低设计成本的目的。半潜式钻井平台是具有高复杂度的工程系统，在其方案设计中要考虑到水动力学、结构力学、动力定位、工程应用以及成本分析等多个学科，非常适合于多学科优化设计的应用。以半潜式钻井平台方案设计为例，介绍了如何使用多学科优化设计来考虑影响平台的多个因素之间的耦合关系，以确定平台的主尺度，来获得最佳的平台总体性能；介绍了应用多学科优化设计时的三个重要关键技术：复杂问题的分解、数学建模和近似模型技术。     

Large-scale network partitioning for decentralized traffic management and other transportation applications

For numerous large-scale engineering and science problems, domain decomposition (DD) has generally been accepted by research communities as among the most attractive methods to obtain solutions efficiently. As a prerequisite for the DD solution process, a large domain must be partitioned into several smaller subdomains, with the key to success (of any DD partitioning algorithm) being the number of system boundary nodes. The lower this number, the more efficiently the subdomains can be processed. Although various transportation researchers have hinted at the use of DD, for example, in intelligent transportation systems-enabled decentralized traffic management, it is assumed that the partition is given. This article presents a simple, efficient, and effective algorithm to decompose a transportation network into a predefined number of interconnected subdomains such that the number of system boundary nodes is small (first priority) and the number of nodes in each subdomain is of similar size (second priority). To assess the effectiveness (in terms of minimizing the number of system boundary nodes) of the proposed Shortest Distance Decomposition Algorithm, it is compared with METIS version 5.1.0, currently among the most widely used graph partitioning algorithms worldwide. Using large-scale, real-world transportation test networks, it was found that the Shortest Distance Decomposition Algorithm is better than METIS in 21 of the 27 examples tested; on average, it provided (approximately) 42.0% of the system boundary nodes (as compared to METIS results) in our large-scale examples.     

A dual decomposition method for sector capacity constrained traffic flow optimization

An aggregate air traffic flow model based on a multicommodity network is used for traffic flow management in the National Airspace System. The problem of minimizing the total travel time of flights in the National Airspace System of the United States, subject to sector capacity constraints, is formulated as an Integer Program. The resulting solution achieves optimal delay control. The Integer Program implemented for the scenarios investigated has billions of variables and constraints. It is relaxed to a Linear Program for computational efficiency. A dual decomposition method is applied to solve the large scale Linear Program in a computationally tractable manner. A rounding algorithm is developed to map the Linear Program solution to a physically acceptable result, and is implemented for the entire continental United States. A 2-h traffic flow management problem is solved with the method.     

Multicomponent decomposition of a time-varying acoustic Doppler signal generated by a passing railway vehicle using Complex Shifted Morlet Wavelets

Complex Shifted Morlet Wavelets (CSMW) present a number of advantages, since the concept of shifting the Morlet wavelet in the frequency domain allow the simultaneous optimal selection of both the wavelet center frequency and the wavelet bandwidth. According to the proposed method, a cluster of CSMW wavelets is used, covering appropriate ranges in the frequency domain. Then, instead of directly processing the instantaneous frequency of each CSMW, an invariance approach is used to indirectly recover the individual harmonic components of the signal. This invariance approach is based actually on the same rotational approach, using the same matrix properties, which consists the core of the well known ESPRIT algorithm. Moreover, the DESFRI (DEtection of Source Frequencies via Rotational Invariance) approach is introduced to support the proposed CSMW method to semi-automated selection of the center frequency of the applied Morlet window. This approach is based on the singular values that are extracted as an intermediate product of the proposed decomposition process. By the application of the method in a multi-component synthetic signal a way to select the critical parameters of the Morlet wavelet, is investigated. The method is further tested on a time-varying acoustic Doppler signal generated by a passing railway vehicle, indicating promising results for the estimation of the variable instantaneous frequency and the multi-component decomposition of it.     

一类虚拟国界预条件多重网络并行算法

对网格方程组作多重网格区域分裂并行计算,以消去法为预条件,讨论了Gauss消去法的迭代性质,指出该迭代法有效改善通常迭代在子域信息传递过程中的障碍性问题,提高多重网格区域分裂并行计算效率。