基于卡口上下文和深度置信网络的车辆轨迹预测模型研究

针对车辆轨迹预测中节点序列的时序特性和实际路网中的空间关联性,该文提出一种基于深度置信网络和SoftMax (DBN-SoftMax)轨迹预测方法.首先,考虑到轨迹在节点集合中的强稀疏性和一般特征学习方法对新特征的泛化能力不足,该文利用深度置信网络(DBN)较强的无监督特征学习能力,达到提取轨迹局部空间特性的目的;然后,针对轨迹的时序特性,该文采用逻辑回归的预测思路,用当前轨迹集在路网特征空间中的线性组合来预测轨迹;最后,结合自然语言处理领域中的词嵌入的思想,基于实际轨迹中节点存在的上下文关系,运用节点的向量集表征了节点间的交通时空关系.实验结果表明该模型不仅能够有效地提取轨迹特征,并且在拓扑结构复杂的路网中也能得到较好的预测结果.     

基于多尺度圆周频率滤波与卷积神经网络的遥感图像飞机目标检测方法研究

针对遥感图像飞机目标检测因目标尺度不一存在漏警、虚警等问题,该文基于遥感图像中飞机目标形状特征和灰度变化特点提出了一种多尺度圆周频率滤波(MSCFF)与卷积神经网络(CNN)相结合的MSCFF+CNN飞机目标自动检测算法.该算法首先采用多尺度圆周频率滤波器滤除遥感图像复杂背景,实现不同尺度飞机目标候选区域提取;然后,通过构建卷积神经网络(CNN)模型实现候选区域有效分类,最终精确确定飞机目标位置.最后,基于获取的真实遥感图像进行目标检测算法实验验证,经统计该算法的飞机目标检测率为94.38%,虚警率为3.76%,实验结果充分验证了该文算法的有效性,该算法可为机场监管、军事侦察等应用提供重要的技术支持.     

基于超像素级卷积神经网络的多聚焦图像融合算法

该文提出了基于超像素级卷积神经网络(sp-CNN)的多聚焦图像融合算法。该方法首先对源图像进行多尺度超像素分割,将获取的超像素输入sp-CNN,并对输出的初始分类映射图进行连通域操作得到初始决策图;然后根据多幅初始决策图的异同获得不确定区域,并利用空间频率对其再分类,得到阶段决策图;最后利用形态学对阶段决策图进行后处理,并根据所得的最终决策图融合图像。该文算法直接利用超像素分割块进行图像融合,其相较以往利用重叠块的融合算法可达到降低时间复杂度的目的,同时可获得较好的融合效果。     

Computing optimal scalings by parametric network algorithms

Asymmetric scaling of a square matrixA 0 is a matrix of the formXAX ^–1 whereX is a nonnegative, nonsingular, diagonal matrix having the same dimension ofA. Anasymmetric scaling of a rectangular matrixB 0 is a matrix of the formXBY ^–1 whereX andY are nonnegative, nonsingular, diagonal matrices having appropriate dimensions. We consider two objectives in selecting a symmetric scaling of a given matrix. The first is to select a scalingA of a given matrixA such that the maximal absolute value of the elements ofA is lesser or equal that of any other corresponding scaling ofA. The second is to select a scalingB of a given matrixB such that the maximal absolute value of ratios of nonzero elements ofB is lesser or equal that of any other corresponding scaling ofB. We also consider the problem of finding an optimal asymmetric scaling under the maximal ratio criterion (the maximal element criterion is, of course, trivial in this case). We show that these problems can be converted to parametric network problems which can be solved by corresponding algorithms.This research was supported by NSF Grant ECS-83-10213.     

Network theory of the nonlinear behaviour of polymer melts. Simple shear flow

Summary The network model developed in a previous paper is applied to the simple shear flow of polymer melts. The constitutive equation obtained consists of two terms. One of them describes the stress due to the network strands which exist at the onset of the deformation, dissociate during the deformation and result in a single integral constitutive equation with a strain dependent damping function. The formulation of the damping function in invariant form seems to be almost impossible.The second normal stress differenceN ₂ of the model is not zero,but has negative values. According to our model this is a consequence of the deformation dependence of the disentanglement process. The theory is compared with experimental data for a LDPE melt. It is found that the model explains the main features of the shear flow behaviour of the LDPE melt investigated preciously.

---

Zusammenfassung Das Netzwerk-Modell, das in einer vorangegangenen Arbeit entwickelt wurde, wird für die einfache Scherströmung von Polymerschelzen angewendet. Die abgeleitete rheologische Zustandsgleichung besteht aus zwei Gliedern. Das erste beschreibt die Spannung-Dehnung-Beziehung der Kettensegmente, die zu Beginn der Deformation existieren und während der Deformation aufgelöst werden. Es hat die Form der einfachen Integralbeziehung mit einer Gedächtnisfunktion. Es ist kaum möglich, die dabei erhaltene Gedächtnisfunktion als Funktion der Invarianten der der Tensoren darzustellen. Die zweite Normal-SpannungsdifferenzN ₂ des Modells ist nicht Null und hat einen negativen Wert. Dies ist nach unserem Modell eine Folge der Deformationsabhängigkeit des Entschlaufungsprozesses. Die Theorie wird mit dem experimentellen Daten für eine LDPE-Schmelze verglichen, wobei sich zeigt, daß das Modell die wesentlichen Merkmale des Scherverhaltens der LDPE-Schmelze gut erklärt.

     

Entrapment and release of sodium polystyrene sulfonate (SPS) from calcium alginate gel beads

The release of sodium polystyrene sulfonate (SPS) from calcium alginate hydrogel beads has been studied. It has been shown that the structure of the cross-linked calcium alginate network is of primary importance in the retention and/or release of the SPS. This has been evidenced by studying the influence of Ca²⁺ concentration, molar masses (M_n) and the ratio of mannuronic acid/guluronic acid components. A minimum in the SPS release is observed in relation with the organization of the network structure. Conditions inducing the organization of a strong gel (e.g. high Ca²⁺ concentration for example) are not always related to a low release. A good control of release is found when a compromise between a well-structured hydrogel and sterical consideration of SPS is reached.     

Cost operator algorithms for the transportation problem

A primal transportation algorithm is devised via post-optimization on the costs of a modified problem. The procedure involves altering the costs corresponding to the basic cells of the initial (primal feasible) solution so that it is dual feasible as well. The altered costs are then successively restored to their true values with appropriate changes in the optimal solution by the application of cell or area cost operators discussed elsewhere. The cell cost operator algorithm converges to optimum within (2T – 1) steps for primal nondegenerate transportation problems and [(2T + 1) min (m, n)] – 1 steps for primal degenerate transportation problems, whereT is the sum of the (integer) warehouse availabilities (also the sum of the (integer) market requirements) andm andn denote the number of warehouses and markets respectively. For the area cost operator algorithm the corresponding bounds on the number of steps areT and (T + 1) min (m, n) respectively.This report was prepared as part of the activities of the Management Sciences Research Group, Carnegie—Mellon University, under Contract N00014-67-A-0314-0007 NR 047-048 with the U.S. Office of Naval Research.     

Influence of silica fillers during the electron irradiation of DGEBA/TETA epoxy resins, part II: Study of the thermomechanical properties

This work describes the influence of silica fillers on the thermomechanical properties of diglycidyl ether of bisphenol A/triethylenetetramine (DGEBA/TETA) epoxy resins during ageing under electron beam irradiation. Whatever be the silica filler (pure micrometric ground and spherical silicas, nanometric silicas and coupling agent treated silicas), the glass transition temperature of the epoxy resins decreases with increasing irradiation dose, meaning that the main effect of the irradiation is chain scission. No influence of the silica fillers has been detected from the changes in the glass transition temperature with the increase in the irradiation dose. The disappearance of the cooperativity of the γ relaxation, the decrease of the α relaxation and the decrease of the elastic modulus at the rubbery plateau observed by dynamic mechanical analyses involve a decrease in the crosslink density of the epoxy resins. The occurrence of chemical reactions between the epoxy resin and the silica surface at high irradiation doses has been shown. Moreover, we show evidence that chemical reactions between the epoxy resin and the silica surface occur at high irradiation dose.     

New double negative and positive temperature coefficients of conductive EPDM rubber TiC ceramic composites

In this paper, we have successfully prepared ethylene-propylene-diene monomer (EPDM)/TiC composites as thermistors, with new double negative and positive temperature coefficients of conductivity (NTCC/PTCC). EPDM composites loaded with 50 phr HAF carbon black and different concentrations of TiC were prepared. This study focuses on the effect of TiC content on the vulcanization process, the network structure and the electrical and thermal properties of EPDM/TiC composites. The effect of TiC on the network structure was evaluated e.g. the curing process, the characteristic time constant during vulcanization, the volume fraction of rubber, gel fraction, interparticle distance between conductive particles, the extent of TiC reinforcement in the rubber matrix and molecular weight between cross-linking through experimental and affine-phantom models. The effects of TiC content on the percolation theory, electrical conductivity, conducting mechanism of conductivity, conducting hysteresis and I-V characteristics were also studied, as well as its TiC on the (NTCC/PTCC), thermoelectric power, dielectric constant and thermal conductivity. Stability and reproducibility of the thermal cycles for heating element applications was tested. Specific heat and the amount of heat transfer by radiation and convection as a function of TiC content was calculated using both the calorimetric technique and a theoretical model. It was proved that TiC improves the network structure, electrical and thermal properties of EPDM composites for practical applications.     

Efficient clustering algorithms for self-organizing wireless sensor networks

Self-organization of wireless sensor networks, which involves network decomposition into connected clusters, is a challenging task because of the limited bandwidth and energy resources available in these networks. In this paper, we make contributions towards improving the efficiency of self-organization in wireless sensor networks. We first present a novel approach for message-efficient clustering, in which nodes allocate local “growth budgets” to neighbors. We introduce two algorithms that make use of this approach. We analyze the message complexity of these algorithms and provide performance results from simulations. The algorithms produce clusters of bounded size and low diameter, using significantly fewer messages than the earlier, commonly used, Expanding Ring approach. Next, we present a new randomized methodology for designing the timers of cluster initiators. This methodology provides a probabilistic guarantee that initiators will not interfere with each other. We derive an upper bound on the expected time for network decomposition that is logarithmic in the number of nodes in the network. We also present a variant that optimistically allows more concurrency among initiators and significantly reduces the network decomposition time. However, it produces slightly more clusters than the first method. Extensive simulations over different topologies confirm the analytical results and demonstrate that our proposed methodology scales to large networks.