基于粒子滤波的多区域辨识性建模跟踪算法

为使跟踪算法中的目标模型不被相同特征的背景像素所干扰,在粒子滤波的框架下引入多区域辨识性建模机制,提出1种新的鲁棒的目标跟踪算法。提出的算法侧重能够辨识前景与背景的最有效信息,把目标物体划分为多个子区域。通过统计子区域与背景类别的类内/类间特性选择出最具信息量的子区域。在选出的目标区域基础上,提出的算法以贝叶斯的方式同时考虑图像信息及空间信息,建立具有辨识性的目标表观模型,并在粒子滤波的更新阶段用以估量目标的所在区域。在一系列具有挑战性的视频序列上的实验结果证明了,在许多复杂场景下,提出的方法与传统的粒子滤波跟踪器相比的鲁棒性以及有效性。     

Linear hybrid simulations of low-frequency fishbone instability driven by energetic passing particles in tokamak plasmas

A linear simulation study of energetic passing particle-driven low-frequency fishbone instability in tokamak plasmas has been carried out using the global kinetic-MHD (magnetohydrodynamics) hybrid code M3D-K. This work is focused on the interaction of energetic passing beam ions andn=1 mode with a monotonic safety factorqprofile andSpecifically, the stability and mode frequency as well as mode structure of then=1 mode are calculated for scans of parameter values of beam ion beta, beam ion injection energy, beam ion orbit width, beam ion beta profile, as well as background plasma beta. The excited modes are identified as a low-frequency fishbone with the corresponding resonance ofwhereis the beam ion toroidal transit frequency andis the beam ion poloidal transit frequency. The simulated mode frequency is approximately proportional to the beam ion injection energy and beam ion orbit width. The mode structure is similar to that of internal kink mode. These simulation results are similar to the analytic theory of Yuet al.     

一种障碍环境下机器人路径规划的蚁群粒子群算法

针对机器人在障碍环境下寻找最优路径问题, 提出了一种障碍环境下机器人路径规划的蚁群粒子群算法.该方法有效地结合了粒子群算法和蚁群算法的优点, 采用栅格法进行环境建模, 利用粒子群算法的快速简洁等特点得到蚁群算法初始信息素分布, 以减少迭代次数, 加快算法的收敛速度; 同时利用蚁群算法之间的可并行性, 采用分布式技术实现蚂蚁之间的并行搜索, 求解精度高等优点, 求精确解. 仿真实验结果证明了该方法的有效性, 是机器人路径规划的一种较好的方法.     

The Effect of Formulation on Radioiodide Thyroid Uptake in the Hyperthyroid Cat

This investigation was designed to compare in vitro dissolution profiles from sodium iodide capsules with radioiodide thyroid uptake in hyperthyroid cats using sodium iodide capsules prepared with a formulation exhibiting a complete release of radioiodide (I-123) in vitro and a formulation with an incomplete release of radioiodide. In vitro dissolution profiles for I-123 sodium iodide capsules with two different formulations were determined using the USP XXIII dissolution test. The two formulations studied in vitro were sodium phosphate dibasic powder with 1% magnesium stearate and calcium phosphate dibasic powder with 3% magnesium stearate. By 20 min after initiation of the dissolution test, over 95% of the I-123 was released from capsules of sodium phosphate dibasic powder. The capsules of calcium phosphate dibasic powder reached 75% at 65 min, with no further release occurring thereafter. There was a statistically significant difference in the dissolution profiles of the two formulations. The thyroid uptake of I-123 from capsules exhibiting complete release and incomplete release of radioiodide was determined in hyperthyroid cats. At 4 hr, the mean percentage thyroid uptake value for sodium phosphate dibasic powder with 1% magnesium stearate (complete release formulation) was 12.0% compared to 9.4% for calcium phosphate dibasic powder with 3% magnesium stearate (incomplete release formulation); at 24 hr, the values were 34.4% compared to 23.7%. The data suggest that the incomplete dissolution profile observed in vitro may correlate with a reduction in the bioavailability of the radioiodide in vivo. However, using the Wilcoxon signed rank test, statistically significant differences did not occur between the complete release formulation and incomplete release formulation at either 4 hr or 24 hr (p >. 05). The results of the in vivo study with five hyperthyroid cats were not conclusive due to the variability in response between individual cats.     

Structural Insights into Iron Ions Accumulation in Dps Nanocage

Dps (DNA-binding protein from starved cells) is well known for the structural protection of bacterial DNA by the formation of highly ordered intracellular assemblies under stress conditions. Moreover, this ferritin-like protein can perform fast oxidation of ferrous ions and subsequently accumulate clusters of ferric ions in its nanocages, thus providing the bacterium with physical and chemical protection. Here, cryo-electron microscopy was used to study the accumulation of iron ions in the nanocage of a Dps protein from Escherichia coli. We demonstrate that Fe²⁺ concentration in the solution and incubation time have an insignificant effect on the volume and the morphology of iron minerals formed in Dps nanocages. However, an increase in the Fe²⁺ level leads to an increase in the proportion of larger clusters and the clusters themselves are composed of discrete ~1–1.5 nm subunits.     

Effect of particle size distribution on green properties during high velocity compaction

Iron powders with two different particle size distributions were compacted by high velocity compaction. The influences of particle size distribution and impact velocity on green properties, including green density, springback, tensile strength and bending strength etc., were studied with scanning electron microscopy (SEM) and a computer controlled universal testing machine. The results show that the particle size distribution and the impact velocity strongly affect its properties. Wider size distribution results in green compact with higher density and better strength. Furthermore, springback of compacts is lower produced by the powder with wider size distribution, especially for radial springback. As impact velocity increases, its green density and green strength gradually increases, but the increasing rate of density decreases gradually. No special relation is found between springback and impact velocity. In addition, the axial springback and the bending strength are higher than the radial springback and the tensile strength, respectively.     

求解连续空间优化问题的扩展粒子蚁群算法

扩展蚁群算法是蚁群算法创始人Dorigo提出的一种用于求解连续空间优化问题的最新蚁群算法,但该算法的收敛速度参数和局部搜索参数取值缺乏理论指导,因此其性能受算法参数影响较大.本文提出一种求解连续空间优化的扩展粒子蚁群算法,将粒子群算法嵌入到扩展蚁群算法中用于在线优化扩展蚁群算法参数,减少了参数人为调整的盲目性.从而改善扩展蚁群算法的寻径行为.通过将本文提出的算法与遗传算法、克隆选择算法、蚁群算法、扩展蚁群算法对5种典型测试函数优化的结果对比表明,本文算法在搜索速度和全局搜索能力方面均优于其它算法.     

稠油冷采用冻胶分散体调驱体系的相互作用机制

目的 旨在有效解决稠油化学冷采过程中降黏剂窜流现象严重、油藏动用效率低的问题,支撑稠油绿色高效开发。方法 基于稠油冷采降黏剂高效降黏洗油与冻胶分散体调剖剂储层调控的协同效应,采用复配方式构建了稠油冷采用冻胶分散体调驱体系,测试其基本性能,并使用界面扩张流变仪以及流变仪,考查了体系的界面流变特性和剪切应力特性。结果 体系由质量分数为0.06%～0.12%的冻胶分散体和质量分数为0.05%～0.15%的降黏剂组成,为粒径均一的低黏流体,能够降低界面张力并乳化稠油,降黏率达到95%以上。体系中降黏剂在油水界面的吸附行为决定了体系的乳化降黏能力,降黏剂通过吸附在冻胶分散体的表面提高了体系的聚结稳定性,并考查了组分含量及油藏条件对以上过程的影响。结论 构建了一种兼具储层调控和高效降黏能力稠油冷采用冻胶分散体调驱体系,探明了体系中各组分间的相互作用机制,为稠油化学冷采提供了技术支持。     

一种新的基于粒子群和模拟退火的聚类算法

提出了一种新的基于粒子群和模拟退火的聚类算法。每个粒子作为聚类问题的一个可行解组成粒子群,粒子的位置由聚类中心向量表示。为避免粒子群陷入局部最优解,结合聚类问题的实际特点,提出了利用模拟退火的概率突跳性的两个解决方案。实验结果表明,新算法增强了全空间的搜索能力,性能优于粒子群算法和传统的K-means算法,具有较好的收敛性,是一种有效的聚类算法。     

Magnetic Field Improvement in End Region of Rectangular Planar DC Magnetron Based on Particle Simulation

For a rectangular planar direct current （DC） magnetron, anomalous target erosion may occur in the curve-out region and inner side of the curved region. One key factor is that the magnetic field in the end region is weaker than that in the straight region, and another important factor may be that there is a circumferential component of the magnetic field in the curved region. Through a calculation of three-dimensional magnetic field for the rectangular magnetron, a magnet structure shimmed by permanent magnet bars and ferromagnetic bars is proposed to solve the above problems. Through a three-dimensional non-self-consistent particle simulation and the Yamamura/Tawara formula, the target erosion profile could be predicted. The simulation results show that for an improved uniformity in magnetic field, the entire target utilization could be much enhanced.