A New Algorithm To Solve Vehicle Routing Problems (VRPs)

Efficient Resource management has a direct influence on business performance and profitability. Vehicle Routing Problems (VRPs) considered in this paper are resource management problems where the aim is to use the limited number of resources to a large number of jobs so that the maximum number of jobs can be completed with minimum cost. A VRP consists of a workforce of maintenance engineers and a set of geographically distributed customers requiring certain services. The problem is complicated by incorporating certain technological and temporal constraints. The objective is to maximize the amount of work done measured in terms of total number of jobs completed and to minimize the total distance travelled by all the engineers. The solution to a VRP is a list of engineers and for each engineer a tour consisting of an ordered list of services to be completed by him under the given constraints. These Problems belong to the class of NP-Complete problems. The stochastic techniques such as Hill-Climbing (HC), Tabu Search (TS), Genetic Algorithms (GA) and Simulated Annealing (SA) are found to be suitable for solving these problems efficiently. It is found empirically that out of these SA gives good results for VRPs. But in some cases it also gives poor quality results. This happens due to not allocating intelligently the unallocated jobs in SA in subsequent iterations. A new algorithm is proposed to solve VRPs in this paper. This is achieved by allocating unallocated jobs intelligently in SA. The proposed algorithm is tested empirically on a number of randomly generated VRPs. Three types of VRPs considered are over resourced, under resourced and critically resourced VRPs. In almost all cases, the proposed algorithm completes a large number of jobs with minimum cost in comparison with SA.     

基于SA-WPSO的遥感图像校正方法

提出一种基于SA-WPSO的遥感图像校正方法。该方法利用多项式模型对图像进行初步几何校正,得到多项式校正系数后,将模拟退火(SA)思想引入粒子群优化(PSO)算法,通过改进的SA-WPSO算法优化多项式校正系数,在此基础上实现图像的几何校正。实验结果证明,与二次多项式及三次多项式校正方法相比,该方法的校正精度更高、鲁棒性更好。     

基于改进模拟退火算法的均匀设计表构建

当 为偶数时,使用好格子点法不能产生因素数较大的设计表,且只能在解空间的子空间内产生设计表,所产生的设计表无法保证最均匀。针对上述问题,引入智能计算方法,包括粒子群优化算法和改进的模拟退火算法。对3种算法的优化性能进行比较,结果表明智能算法可以对较大范围的因素数产生设计表,与好格子点法相比,智能算法能够得到均匀性更好的设计表,并且相对粒子群优化算法,改进模拟退火算法构建的均匀设计表的偏差更小。     

可重构系统中软硬任务划分方法研究

软硬件任务划分是可重构系统开发过程中的重要设计步骤,其划分结果直接影响到可重构系统的性能。目前的软硬件任务划分技术大多只考虑了对应用程序或算法的划分结果,忽略了FPGA在配置和通信时的开销,从而导致实际应用效果不理想。介绍了一种基于性能评估的软硬件任务划分方法,即通过对FPGA计算开销、配置开销、通信开销的预评估测试,结合改进的模拟退火算法得出可重构系统中的软硬任务划分结果。实验结果表明,该划分方法具有较好的划分效果和算法收敛速度。     

供应链订单分配优化模型及其模拟退火算法

受4M1E(人、机、料、法、环)因素的随机波动影响,产品的制造过程通常是不完美的,从而产生不良产品.针对已有研究多忽略不良产品的特点,建立了更加符合实际需求的订单分配多目标混合整数规划模型,其优化目标为最小化交易成本、采购成本、不良产品数量、产品延迟交付数量,以及最大化供应商信誉评价.考虑到模型求解的复杂度,设计了一种模拟退火算法,并结合启发式规则避免了大量非法初始解与邻点解的出现.实验算例表明所建立的模型能够反映订单分配过程中的产品缺陷现象,其算法能够在允许的运算时间内获得稳定的满意解,并且随着算例规模的增大,其计算时间与优化结果均优于LINGO软件.     

Capacitive and textural characteristics of hydrous manganese oxide prepared by anodic deposition

The amorphous hydrous manganese oxide (denoted as a-MnO_x·nH₂O) was anodically deposited from the MnSO₄ solutions of various pH values. The capacitive characteristics and stability of this oxide without and with annealing in air for 2 h up to 400 °C were systematically investigated in aqueous electrolytes through means of cyclic voltammetry (CV) and the constant-current charge-discharge method. The redox properties of a-MnO_x·nH₂O were strongly affected by the electrolytes employed and this oxide exhibited ideally capacitive behavior in 0.1 M Na₂SO₄ and 0.3 M KCl. The stability of this amorphous hydrous oxide was enhanced by the annealing treatment while its capacitance was gradually decreased with increasing the annealing temperature. The amorphous structure and surface morphologies of a-MnO_x·nH₂O with annealing at different temperatures were, respectively, examined in terms of the X-ray diffraction (XRD) patterns and scanning electron microscopic (SEM) photographs. The oxidation states of these a-MnO_x·nH₂O deposits were studied by X-ray photoelectron spectroscopy (XPS).     

基于混合t聚类的鲁棒非刚体点匹配

针对非刚体形状控制点匹配中存在的对应性和有偏性问题,提出一个基于混合t聚类的鲁棒点匹配模型。该模型结合了混合t分布聚类和薄片样条匹配各自的优点,混合t聚类充分利用t分布的小样本和较大尾迹优势取得局部聚类的鲁棒性,并可方便地引入新的成员分布概率建模局外点,解决全局的鲁棒性;薄片样条能完全分解非刚体映射的仿射和非仿射变换。实验证明,利用确定性退火和EM(期望最大化)迭代技术,对模型中的鲁棒聚类和非刚体点匹配进行联合优化,可以有效解决非刚体点匹配的对应性和有偏性问题;通过在退火过程中应用一个模型选择准则,可以进一步提高匹配的精度,对比结果证实了该模型的鲁棒性。     

Deep levels in GaInAs grown by molecular beam epitaxy and their concentration reduction with annealing treatment

X-ray diffraction (XRD), current–voltage (IV), capacitance–voltage (CV), deep-level transient Fourier spectroscopy (DLTFS) and isothermal transient spectroscopy (ITS) techniques are used to investigate the thermal annealing behaviour of three deep levels in Ga_0.986In_0.014As heavily doped with Si (6.8 × 10¹⁷ cm⁻³) grown by molecular beam epitaxy (MBE). The thermal annealing was performed at 625 °C, 650 °C, 675 °C, 700 °C and 750 °C for 5 min. XRD study shows good structural quality of the samples and yields an In composition of 1.4%. Two main electron traps are detected by DLTFS and ITS around 280 K, with activation energies of 0.58 eV and 0.57 eV, capture cross sections of 9 × 10⁻¹⁵ cm² and 8.6 × 10⁻¹⁴ cm² and densities of 2.8 × 10¹⁶ cm⁻³ and 9.6 × 10¹⁵ cm⁻³, respectively. They appear overlapped and as a single peak, which divides into two smaller peaks after annealing at 625 °C for 5 min.

Annealing at higher temperatures further reduces the trap concentrations. A secondary electron trap is found at 150 K with an activation energy of 0.274 eV, a capture cross section of 8.64 × 10⁻¹⁵ cm² and a density of 1.38 × 10¹⁵ cm⁻³. The concentration of this trap level is also decreased by thermal annealing.     

Effect of annealing on the characteristics of Au layers grown on the high-temperature deposited Ni50Fe50 layers

80 nm-thick Ni₅₀Fe₅₀ layers were sputter-deposited on glass substrates at 400 °C and then Au layers were sputter-deposited on the Ni₅₀Fe₅₀ layers. The Au/Ni₅₀Fe₅₀ bilayer films were annealed in a vacuum of 5×10⁻⁴ Pa from 250 to 450 °C for 30 min or 90 min. The characteristics of the Au layers were studied by Auger electron spectroscopy, field emission scanning electron microscopy, X-ray diffraction and a four-point probe technique. When the annealing temperature reaches 450 °C, Fe and Ni atoms diffuse markedly into the Au layer and the Fe content is more than the Ni content. When the annealing temperature is lower than 450 °C, the grain size of the Au layers does not change markedly with annealing temperature. However, as the annealing temperature reaches 450 °C, the annealing promotes the grain growth of the Au layer. As the annealing temperature exceeds 300 °C, the resistivity of the bilayer films increases with increasing annealing temperature. The diffusion of Fe and Ni atoms into the Au layer results in an increase in the resistivity of the annealed bilayer film. Large numbers of Fe and Ni atoms diffusing into the Au layer of the annealed Au/Ni₅₀Fe₅₀ bilayer film lead to a significant decrease in the lattice constant of the Au layer.     

Evolution of internal stress and microstructure in Ti50Cu50 alloy films: influence of substrate temperature and composition

The growth stress of Ti₅₀Cu₅₀ alloy films on alumina substrates was measured in situ under ultrahigh vacuum conditions with a cantilever beam technique as a function of substrate temperature and variation of copper concentration. The growth stress of films deposited at low temperature is interpreted to indicate the growth of amorphous respectively nanocrystalline films. At substrate temperatures above 300 °C, a novel tensile stress contribution is interpreted to be due to the formation of a preferentially oriented film of metastable copper vacancy superstructured γ-TiCu. Finally, at substrate temperatures above 365 °C, a polycrystalline γ-TiCu film is formed.Annealing of low-temperature alloy films gives rise to irreversible tensile stress changes indicating restructuring of the films. The stress change measured when annealing a superstructured γ-TiCu film is compressive and is assigned to the transformation of the metastable superstructure to polycrystalline γ-TiCu. The magnitude of this compressive stress contribution is strongly dependent on the copper concentration, i.e., copper vacancy concentration.