改进萤火虫算法及其收敛性分析

萤火虫算法因具有结构简单、控制参数少、易于实现的特点而得到广泛的关注和应用,但其易陷入局部最优导致过早收敛,从而影响寻优精度。针对这一问题,本文在位置更新规则中加入随机扰动因子,并剔除了冗余的随机项,以提高算法搜索能力;引入位置置换变异和差分进化算法中的最优变异策略,在保持种群多样性的同时,增强算法跳出局部最优的能力。采用马尔可夫过程证明了算法以概率1收敛到全局最优。利用基准函数和装箱问题对算法进行仿真测试,结果表明,改进后的算法能够有效跳出局部最优,对给出的所有问题均能找到理论最优解,寻优精度和成功率有明显提升。     

Estimates of true birth order for Scotland, 1945-1999

Over the last three decades fertility in Scotland, as measured by the total fertility rate (TFR), has moved from being higher than in England and Wales, to being lower. The annual number of births in Scotland has declined so that in the mid-1990s low fertility became the main driver of the overall population decline that Scotland has been experiencing since 1974. Analysis of fertility by birth order is instrumental in gaining an understanding of past and future fertility trends. Until the rise in births outside marriage in the 1980s data from registration could be used as a proxy for true birth order. However, because birth order is not collected for births outside marriage true birth order now has to be estimated. This article presents the first official estimates of true birth order for Scotland. The construction of these estimates based on a modified version of the method used for England and Wales is discussed. This article also presents analysis relating births by true birth order estimate to the population of women by parity on a cohort basis, and makes comparisons with England and Wales.     

协同粒子群-模拟退火算法求解VRPSPD问题

研究逆向物流车辆路径(VRPSPD)问题,建立了VRPSPD问题的混合整数规划模型.利用启发式算法的特点构造求解VRPSPD问题的一种协同PSO_SA算法,设计了该算法初始种群的编码规则、信息交换策略、2-opt邻域解生成策略和SA算法中的冷却进度表规则.实验过程以典型算例为例进行了实验,并对重要参数进行了分析.实验结果表明,该算法对于求解VRPSPD 问题,可以有效提高车辆的负载使用率,避免因负载波动和最大负载能力约束而增加车辆总行程,在可以接受的迭代次数限制内可以收敛到满意解.     

一种自适应多亲遗传算法及其性能分析

定义了一种度量多样性的指标,来自适应地调整基于实数编码的多亲遗传算法中交叉概率和变异概率这两个参数。提出了一种基于实数编码的自适应多亲遗传算法,该算法能自适应地调整其参数,且在求解优化问题的过程中,能克服早熟收敛的现象,提高搜索能力,加速收敛速度。同时将该算法用来求解多峰值测试函数,实验结果验证了算法的有效性。     

一种求解资源受限工程调度问题的遗传算法

针对资源受限工程调度问题的特点，设计了一种改进的遗传算法，对算法的编码方式和解码规则，适值函数，遗传算法、初始种群的产生方法进行了详细的说明，利用被普遍应用的标准问题对该算法进行了大量的测试，分析了算法参数和问题参数对算法效果的影响，取得了令人满意的结果。     

求解有硬时间窗车辆路径问题的改进遗传算法

针对军事运输中有硬时间窗的车辆路径问题(vehicle routing problem with hard time windows, VRPHTW),结合混合交叉运算、改进变异运算和精英保留策略,以所有车辆的配送总时间最少为目标,设计了改进遗传算法。借鉴贪婪思想,提高了初始种群的优越性;构造了迭代种群的入口矩阵和出口矩阵,并以此为基础提出改进交叉算子,期间引入前向插入法设计了混合交叉运算,加快了种群的寻优速度;同时提出改进变异算子,增加了种群的多样性。实验结果表明,改进遗传算法较之基本算法有着更快的收敛速度和更优的收敛效果。     

基于文化克隆选择算法的梯级水电站联合优化调度

为求解梯级水电站联合优化调度问题,提出文化克隆选择算法（CCSA）。CCSA将克隆选择算法（CSA）嵌入文化算法（CA）框架,并根据克隆选择算法的特点,重新定义了文化算法信念空间的4种知识结构,进而利用这些知识结构指导克隆选择算法的演化过程,避免了高频变异对收敛速度的不利影响,从而提高了算法的收敛速度和搜索效率。函数仿真测试结果表明CCSA在继承CSA多样性好、不易早熟特点的基础上,收敛速度也有进一步提高。将CCSA应用于梯级水电站联合优化调度问题的求解,取得满意效果,为求解该问题提供了一种新的可行方法。

Abstract：

A novel optimization approach-cultured clone select algorithm （CCSA） was proposed to solve optimal dispatch problem of cascade hydroelectric stations.CCSA used cultural algorithm （CA） as its framework and clone select algorithm （CSA） in its population space.Considering the characteristics of CSA,CCSA redefined four knowledge structures in belief space and used these structures to guide the evolution process of CSA.By doing this,CCSA abated the adverse affect caused by high-frequency to convergence rate,thus it converged faster without destroying diversity.CCSA is first tested by several test functions,it is found that CCSA can avoid premature convergence effectively and has fast convergence rate.Then CCSA is applied to an optimal dispatch problem of cascade hydroelectric stations,the results show that it is effective and can be an alternative for this problem.     

Resource allocation with minimum transmit power in multicast OFDM systems

To minimize the total transmit power for multicast service in an orthogonal frequency division multiplexing (OFDM) downlink system, resource allocation algorithms that adaptively allocate subcarriers and bits are proposed. The proposed algorithms select users with good channel conditions for each subcarrier to reduce the transmit power, while guaranteeing each user’s instantaneous minimum rate requirement. The resource allocation problem is first formulated as an integer programming (IP) problem, and then, a full search algorithm that achieves an optimal solution is presented. To reduce the computation load, a suboptimal algorithm is proposed. This suboptimal algorithm decouples the joint resource allocation problem by separating subcarrier and bit allocation. Greedy-like algorithms are employed in both procedures. Simulation results illustrate that the proposed algorithms can significantly reduce the transmit power compared with the conventional multicast approach and the performance of the suboptimal algorithm is close to the optimum.     

考虑技术突破与技术升级的能力规划问题研究

技术进步与市场需求不确定性,都是制造业能力规划问题需要面对的最重要因素.以往研究仅考虑技术突破而没有考虑技术升级,较多考虑能力扩张而较少考虑能力更替,本文首次提出了一种同时考虑技术突破和升级的设备采购与替换的集成能力规划模型.在需求和技术进步双源不确定性条件下,采用Scenario方法建立了多计划期能力规划的集成决策模型,探求能力扩张、替换、维护的期望总成本最低.面对该决策问题所建立的非线性混合整数规划模型,本文为该NP complete问题设计了一种基于遗传算法框架的启发式求解算法,即通过增加变换操作的方法将该问题转换为可以随机进化求解扩张方案最优化求解替换方案的一个等价问题.在种群初始化过程中采取仅选择能力扩张决策进行染色体部分编码策略,然后用经典优化方法针对每个个体精确求解最优能力替换决策,将扩张和替换整体成本作为个体适应度参与个体评价与种群进化.实验结果表明,技术升级在不同需求变化情景下都可以有效降低能力规划的成本,且本文提出的启发式算法对于求解此类规划问题具有很好的收敛稳定性和性能稳定性.     

The changing demographic picture of the UK: national statistician's annual article on the population

The population of the UK is currently growing at its fastest rate since the 1960s, increasing by two and a half per cent between mid-2001 and mid-2006. While life expectancy continues to increase, fertility rates have also been increasing in the last five years and are currently at their highest level since 1980. In addition, international migration has led to the UK population growing by an average of 500 people per day over the last five years. The population is also becoming increasingly diverse and mobile, and these factors make it increasingly challenging to measure population change accurately. This is the first of a series of annual reports on the population of the UK; these reports will provide an overview of the latest statistics on the population and will also focus on one specific topic - for this report the topic is fertility and, in particular, the impact of migration on fertility, but different topics will be covered in future years. The reports will also highlight the key strands of work being taken forward within the National Statistics Centre for Demography in order to improve UK population statistics. More detailed information on the populations of Wales, Scotland and Northern Ireland are available in the annual demographic reports produced by the respective devolved administrations.'     

遗传算法平均截止代数和成功率与种群规模之间的关系

遗传算法的搜索效率在很大程度上取决于运行参数,如种群规模M、编码串长度l、交叉概率Pc、变异概率Pm、终止代数T、适应度函数f(M)等的设置.利用简单遗传算法和统计分析方法,通过对四个典型数学结构进行全局最优解搜索,得到了遗传算法平均进化截止代数和成功率R与种群规模M之间的关系,给出了相应的拟合公式.讨论了平均进化截止代数和成功率与种群规模和极值个数之间的关系.这种研究对于合理确定种群规模和进化终止代数,正确估计搜索到的满意解的最优程度等具有重要意义.     

基于粒子群优化的卡尔曼滤波去耦算法

针对相控阵雷达导引头由于前向通道增益和波束控制增益刻度尺度不同引起的扩展卡尔曼滤波(extended Kalman filter, EKF)去耦中误差量过大的问题,提出了基于粒子群优化的EKF去耦算法。采用了最小均方差为适应度函数,对两个增益参数进行组合优化,然后通过建立EKF的系统模型,推导了提取的视线角速率与增益参数之间的关系,使得滤波后的估计值为最优的后验估计。最后,通过仿真表明该算法可以很好地解决误差量过大的问题,并验证了所提算法在相控阵雷达导引头去耦和视线角速率提取中的有效性。     

The effect of changes in timing of childbearing on measuring fertility in England and Wales

Changes in the ages at which women give birth to their children mean that fertility measured at a particular point in time (period) may not be a good representation of the ultimate fertility of those women. The common measure of period fertility is the total fertility rate, which in 2001 has fallen to the lowest level since records began in England and Wales. This article presents various methods that have been proposed to adjust period fertility data to take account of changes in the timing of childbearing, applied to England and Wales data. The article concludes that while these adjustment methods provide useful insights, for example, that the total fertility rate has underestimated period quantum fertility since the 1970s, the measures produced are difficult to interpret. This is in part because the concept they are trying to measure, period quantum is itself imprecise. The adjustments do not necessarily provide a reliable indicator of underlying cohort fertility.     

基于一因多效的遗传算法研究

在自然界中存在着"一因多效"的现象,即某一基因可影响生物的许多性状的生长发育。模仿这一生物现象,提出了基于一因多效的遗传算法,建立了基因型与表现型之间的一对多的非线性关系,打破了遗传算法中惯用的一一对应关系。并阐述了一因多效遗传算法有利于维护问题空间的多样性,从而可采用小规模种群来提高算法的运行效率。以求解Rosenbrock函数的全局最优解为例,通过与传统遗传算法的对比实验,证实了一因多效遗传算法的可行性。     

基于改进差分进化算法的水文模型参数多目标优选研究

在差分进化算法的基础上, 受马尔可夫链蒙特卡罗方法的启发, 建立了differential evolution adaptive metropolis (DREAM)算法. DREAM 算法融合了马尔可夫链蒙特卡罗方法和差分进化算法的优势, 较好地解决了马尔可夫链蒙特卡罗方法中搜索步长的恰当取值以及搜索方向的准确定位问题, 并能有效解决差分进化算法的群体多样性和收敛速度问题. 在 DREAM 算法基础上, 引入多目标优化思想, 提出了一种基于改进适应度分配策略和外部存档方案的多目标 DREAM 算法, 并应用于岷江流域 CMD-3PAR 降雨-径流模型参数优选研究. 结果表明: 多目标DREAM算法能够找到一组范围宽广、分布均匀且数量充足的 Pareto 最优解供决策者评价优选.     

基于合理膳食结构的2020-2050年我国食物用粮需求测算

测算我国未来粮食需求量是制定我国粮食安全相关政策措施的基础.目前我国城镇化水平加快,老龄化问题突出,完全二孩生育政策刚刚实施,正处于人口结构转型的关键期,而人口结构变动直接影响食物用粮需求.本文从满足居民营养健康标准的食物需求视角下,考虑人口年龄、性别和城乡结构,采用标准人消费系数法,测算了单独二孩政策、完全二孩政策和完全放开生育政策情景下2020-2050年我国食物用粮（包括口粮与饲料粮）的需求,并进行对比分析.结果表明：在完全二孩政策情景下,我国食物用粮需求将在2030年达到峰值39263.1万吨,比通常采用人均方法的计算结果低10061.0万吨;因城镇化率的提高将使2020、2030与2050年食物用粮需求分别增加约406.7万吨、1142.5万吨与1553.8万吨;生育政策调整所带来的人口规模与结构变化并不是2020-2050年我国粮食安全的主要影响因素.建议积极引导和鼓励生育,在适当时间进一步完全放开生育;同时采取有效措施将粮食损耗与浪费降到最低.     

基于FPGA的实时整数霍夫变换

为了解决霍夫变换计算量大、难以实时实现及峰值检测不准确的问题,提出一种适合在FPGA上实现的基于两阶段快速搜索算法的改进整数霍夫变换,并将其在FPGA中实现。首先,设计了一种基于该改进霍夫变换算法的流水线结构,能够在单个时钟周期内完成霍夫空间的参数计算和累加;接着,设定3个参数阈值,以寻找霍夫子空间中的局部极值;最后,确定全局极值,实现直线参数的提取。采用实际图像在FPGA中进行实验验证,结果表明提出的算法占用硬件资源较少,其准确率达到93%以上。     

多目标置换流水车间调度的混沌杂草优化算法

针对最小化最大完工时间,总流程时间及总延迟时间的多目标置换流水车间调度问题,提出一种改进的混沌杂草优化算法,该算法采用基于熵值权重的灰熵关联度适应值分配策略,引入快速非支配排序法生成外部档案,并将进化种群的更新和最优位置的混沌搜索相结合,用于维护外部档案,提升算法的寻优性能.通过与NSGA-Ⅱ算法进行OR-Library典型测试算例的对比实验,验证该算法的有效性.     

QoS multicast routing algorithm based on GA

1 .INTRODUCTIONTheprovisionofquality of service (QoS) guaranteesisofutmostimportanceforthedevelopmentofthemulticastservices .Multicastroutinghascontinuedtobeaveryimportantresearchissueintheareasofnet worksanddistributedsystems.Ithasattractedtheinterestsof…     

废弃物收运网络周期性选址-路径问题多目标优化模型及算法

废弃物处理设施是典型的"Semi-desirable"设施, 其选址相关的决策需要考虑成本、负效应等多个相互冲突的目标. 废弃物中转站的选址和路径优化是一类周期性选址-路径问题(PLRP), 目前缺乏对这类问题的多目标优化的研究. 以多目标城市废弃物收运网络周期性选址-路径问题为研究对象, 提出基于扩散距离的负效应测度方法; 建立考虑负效应和成本的双目标混合整数规划模型; 设计DRECWA-LS算法, 结合启发式优化策略, 求解收运计划对应的周期性选址-路径问题, 并采用扩展邻域搜索策略的多目标进化算法实现对可行收运计划的深度搜索, 同步求解收运计划、中转站选址-分配及车辆路径三个问题. 算例结果显示: 算法能成功求解较大规模问题, 具有较高的求解质量和运算效率.