基于亲缘选择的遗传算法

针对传统遗传算法容易陷入局部最优解的缺陷,借鉴生物中亲缘选择的思想,提出基于亲缘选择的遗传算法。该算法构造新选择算子,通过按亲缘关系放弃一个解而获得另一个解来保证算法在最优解的领域内的有效搜索,提高遗传算法对全局最优解的搜索能力和收敛速度。仿真结果表明,该算法正确有效,性能优于现有的传统算法。     

基于遗传算子的改进粒子群优化算法

为了克服PSO算法容易陷入局部最优的缺陷,提出一种基于遗传算子的改进PSO算法。该算法借鉴常规的遗传算法中的选择交叉操作,在优化搜索过程中更新粒子的位置时,进行交叉操作,可以扩大全局搜索范围,避免局部最优,提高粒子的多样性。对改进后的算法使用几个典型的测试函数进行了仿真实验,实验结果表明,相比于标准PSO该算法的全局搜索能力和收敛精度都有较大提高,有效地改善了优化性能。     

动态串行机制多元宇宙优化算法

为优化多元宇宙算法求解函数最优值的性能,提出一种改进搜索机制的全局优化多元宇宙算法(G-MVO).针对标准算法存在单一搜索机制导致算法易陷入局部最优以及过早收敛的缺陷,提出三种学习策略来增强算法性能,通过多策略交互协作降低算法复杂度并提高求解精度,设计自适应参数动态选择最佳策略,全局优化算法性能.为验证算法的有效性,算法在不同维度的八个基准函数上进行仿真实验.结果表明,该算法表现出更佳的求解精度以及收敛速度.     

基于改进全局和声搜索算法LSSVM的短期电力负荷预测

针对全局和声搜索算法（GHS）存在的缺陷,提出改进全局和声搜索算法（IGHS）。该算法通过扩大最优和声搜索区域,并在搜索过程中引入受和声库影响的微调变量,从而增强了算法跳出局部极小值束缚的能力。将该改进算法应用于电力负荷预测中,提出基于改进全局和声搜索算法最小二乘支持向量机（IGHS-LSSVM）的负荷预测方法。利用某电力公司的历史数据进行仿真,结果表明方法具有较高的预测精度。     

基于蚁群算法的QoS最佳路由选择问题的研究

针对当前通信网络中存在的服务质量问题提出了满足QoS约束的最佳路由选择算法——蚁群算法,该算法是一种新型的进化算法。与其它进化算法一样存在搜索时间过长和易于陷入局部最优解等缺陷。提出了一种通过动态地调整信息素的挥发度和采用最优、最差路径信息索全局更新策略来克服上述缺陷。改进型算法能较快地收敛到全局最优解。仿真结果表明上述方法是可行和有效的。     

改进的重叠蚁群优化算法

针对蚁群优化算法在进行全局最优解搜索时容易陷入局部最优解和收敛速度缓慢等缺陷,提出了一种有效求解全局最优解搜索问题的重叠蚁群优化算法。该算法通过设置多个重叠的蚁群系统,并对每一个蚁群初始化不同的参数,之后在蚁群之间进行信息素的动态学习,增强了不同蚁群对最优解的开采能力,避免了算法出现早熟现象。仿真实验结果表明,重叠蚁群优化算法在避免陷入局部最优解方面具有良好的效果,是一种提高蚁群算法性能的有效的改进算法。     

禁忌搜索算法与蚁群算法的混合策略在二次分配问题上的应用

二次分配问题是一个NP-hard问题,它在线路板设计、布局问题以及打字机键盘的设计等现实生活中有许多的应用.使用基本蚁群算法进行搜索时,其全局优化性能的优劣在很大程度上与蒸发系数的选择有关,若选择不合适,易使算法陷入局部最优.为此,本文提出一种新的算法,即将基本蚁群算法与禁忌搜索策略相结合来求解二次分配问题,设计出具体的算法模型,并对标准问题库中的具体实例进行测试,实验结果证实新方法的有效性.     

全局调距和声特征选择算法

特征选择技术能有效解决维数灾难问题，许多搜索策略已经被应用到特征选择问题中。针对和声特征选择算法搜索能力低下的问题，提出了一种基于全局自适应调距的和声特征选择算法（HSFS-GPA）。将特征集的距离定义引入到特征选择问题中，在算法搜索过程中结合全局信息对随机产生的新和声进行调整，以一定概率减小候选和声与当前最优和声的距离来加快算法搜索速度，或减少候选和声与最差和声的距离以避免陷入局部最优；同时，采用竞争选择方案随时更新和声库全局信息，改进和声库的更新机制提高算法搜索质量。将HSFS-GPA与原始和声特征选择算法、粒子群算法和遗传算法进行对比实验，HSFS-GPA所选特征子集的大小比原始和声算法减少15%，子集评价值平均提高到0.98。实验结果表明，HSFS-GPA能在相同的条件下搜索到更优质的特征子集。     

基于云模型的实数编码量子进化算法

针对量子进化算法易陷入局部最优和求解精度不高的缺点,利用云模型具有随机性和稳定倾向性的特点,提出了一种基于云模型的实数编码量子进化算法。该算法利用单维云变异进行全局快速搜索,利用多维云进化增强算法局部搜索能力,探索全局最优解。依据算法的进化过程动态调整搜索范围并复位染色体,可以加提高敛速度,并防止陷入局部最优。仿真结果表明,该算法搜索精度和效率得到提高,适合求解复杂函数优化问题。     

免疫遗传蚁群融合算法

提出了一种融合蚁群系统、免疫算法和遗传算法的混合算法。将免疫算法和遗传算法引入到每次蚁群迭代的过程中,利用免疫算法的局部优化能力和遗传算法的全局搜索能力,来提高蚁群系统的收敛速度。该算法通过遗传算法的选择、交叉、变异操作和免疫算法的自适应疫苗接种操作,有效地解决了蚁群系统的易陷入局部最优和易退化的缺点。通过对旅行商问题的仿真实验表明该算法具有非常好的收敛速度和全局最优解的搜索能力。     

A decentralized coordination method for optimal load redistribution in heterogeneous service grids

The optimal load redistribution problem is solved in this paper for heterogeneous non‐dedicated service grids in a decentralized way. A coordination policy is proposed to make networked servers reach their optimal generic task acceptance rates in order to minimize the average service time of all the generic tasks in a grid. Autonomous servers networked in the grid only need to coordinate with their neighbors iteratively, and their optimal generic acceptance rates are reached by task migration among them. The design scheme of the policy is introduced, and the convergence properties and the implementation aspects of the coordination system are discussed in detail in this paper. A set of computer simulations have been conducted, validating the effectiveness of the proposed approach. Copyright © 2010 John Wiley & Sons, Ltd.     

遗传算法的改进与应用

遗传算法是一种借鉴生物界自然选择和自然遗传机制的随机搜索算法,它与传统的算法不同。大多数古典的优化算法是基于一个单一的度量函数(评估函数)的梯度或较高次统计,以产生一个确定性的试验解序列;遗传算法不依赖于梯度信息,而是通过模拟自然进化过程来搜索最优解。该文针对传统遗传算法的缺陷,提出了一些新的改进思路,即从搜索技术和遗传算子等的角度来改进遗传算法。     

基于云环境下新产品开发团队优选研究

针对云环境下新产品开发团队优选问题,充分考虑团队的研发能力、协调能力和服务质量,建立了包含知识相似度评价模型、协同效应评价模型和服务质量评价模型的综合评价模型;对算法的适应度函数和搜索方式加以改进,提出了改进的人工蜂群算法,对所建立的模型进行求解,为服务需求方选择出最优团队组合;最后,以自动引导运输车（AGV）的新产品开发为例,通过模型求解和算法对比,验证了所提方法的可行性和有效性。     

The rough set based approach to generic routing problems: case of reverse logistics supplier selection

In recent years, Reverse Logistics (RL) has been touted as one of the strategies of improving organization performance and generating a competitive advantage. In RL, the generic routing problem has become a focus since it provides a great flexibility in modeling, e.g., selection of suppliers by using a node as a supplier candidate in a network. To date, complicated networks make decision makers hard to search a desired routine. In addition, the traditional network defines and resolves such a problem only at one soot. The solution cannot be acquired from multiple perspectives like minimal cost, minimal delivery time, maximal reliability, and optimal “3Rs”—reduce, reuse, and recycle. In this study, rough set theory is applied to reduce complexity of the RL data sets and induct decision rules. Through incorporating the decision rules, the generic label correcting algorithm is used to solve generic routing problems by integrating various operators and comparators in the GLC algorithm. Consequently, the desired RL suppliers are selected.     

Architectural mechanisms for dynamic changes of behavior selection strategies in behavior-based systems

Behavior selection is typically a "built-in" feature of behavior-based architectures and hence, not amenable to change. There are, however, circumstances where changing behavior selection strategies is useful and can lead to better performance. In this paper, we demonstrate that such dynamic changes of behavior selection mechanisms are beneficial in several circumstances. We first categorize existing behavior selection mechanisms along three dimensions and then discuss seven possible circumstances where dynamically switching among them can be beneficial. Using the agent architecture framework activation, priority, observer, and component (APOC), we show how instances of all (nonempty) categories can be captured and how additional architectural mechanisms can be added to allow for dynamic switching among them. In particular, we propose a generic architecture for dynamic behavior selection, which can integrate existing behavior selection mechanisms in a unified way. Based on this generic architecture, we then verify that dynamic behavior selection is beneficial in the seven cases by defining architectures for simulated and robotic agents and performing experiments with them. The quantitative and qualitative analyzes of the results obtained from extensive simulation studies and experimental runs with robots verify the utility of the proposed mechanisms.     

降低OFDM立方度量的最优限幅滤波算法及神经网络实现

正交频分复用(OFDM)信号的一个主要缺点是信号包络波动过大。峰均功率比是常用的度量OFDM信号包络波动大小的指标,而近期研究表明立方度量可以更加准确地度量OFDM信号包络波动。传统限幅滤波技术可以有效降低立方度量,但其滤波设计并不能保证处理后的信号性能达到最优。针对这一问题,提出了一种最优的限幅滤波设计方案来降低立方度量,其关键思想是考虑滤波操作对信号带内、带外部分的影响,将滤波器设计建模为一个优化问题,通过求解得到最优的滤波器,并与限幅操作结合降低立方度量。由于优化问题的求解复杂度较高,还提出了一种基于深度神经网络的最优限幅滤波实现方案。仿真结果表明,所提出的最优限幅滤波算法及其神经网络实现方案性能相当,但后者的复杂度要低得多。与其它的已知算法相比,新提出的算法及其神经网络实现方案的性能都具有明显的优势。     

连续型2维图形矩阵码格式设计及掩模方案

2维条码目前已广泛应用于各个领域,2维条码的编码过程中有一个重要步骤称为掩模调制,其目的是为了限制生成图像各区域的平均灰度,以提高码字的易印制、易识别能力。但现有的掩模方法大多基于感性,掩模模式有限并缺乏规范的方法支持及客观的评估手段,难以在各种应用条件下取得最佳效果。本文的研究目的是对能够对2维条码的掩模进行一种客观的数学上的评估。本文结合作者自行提出的一种可承载不确定容量信息的连续型2维条码,从统计分布模型角度对掩模技术进行研究,通过一种归于正态分布的数学评估模型,最终引导出一种比传统方法更为客观、规范、可最优化的掩模方法,该技术还可以应用于QR码等2维编码规则中。     

An improved code selection algorithm for fault prediction

The goal of this study is to present an improved code selection algorithm (BCSA) for fault prediction. The contributions mainly contain three parts. The first part is on the extension of the horizontal input in the code selection algorithm (CSA). We propose that the horizontal input is also the prediction for the next coming event, not only for recalling. Thus, BCSA is able to recall and predict alternately. The second part is on the extension of the generic minicolumnar function. We propose that the function of a minicolumn is to be a k-winner-take-all competitive module (CM) and all active cells (the overall input is 1) should be chosen as winners within a CM. The third part is on the improvement of the competition mechanism. In BCSA, the winners are directly chosen with only one round competition. Thus, computing the input’s similarity G is unnecessary. BCSA is applied to analyze the disaster of the space shuttle Challenger which is a well-known example of fault prediction. Compared to other methods, the result of BCSA is specific, robust and independent of the parameters.     

面向特征选择问题的协同演化方法

特征选择技术是机器学习和数据挖掘任务的关键预处理技术。传统贪婪式特征选择方法仅考虑本轮最佳特征,从而导致获取的特征子集仅为局部最优,无法获得最优或者近似最优的特征集合。进化搜索方式则有效地对特征空间进行搜索,然而不同的进化算法在搜索过程中存在自身的局限。本文吸取遗传算法（GA）和粒子群优化算法（PSO）的进化优势,以信息熵度量为评价,通过协同演化的方式获取最终特征子集。并提出适用于特征选择问题特有的比特率交叉算子和信息交换策略。实验结果显示,遗传算法和粒子群协同进化（GA-PSO）在进化搜索特征子集的能力和具体分类学习任务上都优于单独的演化搜索方式。进化搜索提供的组合判断能力优于贪婪式特征选择方法。     

Designing intelligent agents to support long-term partnership in two echelon e-Supply Networks

Realizing the dynamic nature of information flow and the conflicting objectives of members play vital role in effective design of e-Supply Networks (e-SN). While there are some research in the SN literature proposing different dynamic and intelligent coordination mechanisms, the impact of the proper definition of data structure and long-term relationship in modeling both coordination and negotiation mechanisms have not been addressed deeply. In this paper, we propose three overlapping services including intelligent matching of partners, proposal generations, and long-term contract management. The process begins with the selection of qualified partners based on the similarity of users profiles in a multidimensional space defined by network attributes. Then, a coordination mechanism for long-term agreements is proposed such that the generated proposals in e-SN encourage buyers to reveal their demand in advance. The proposed mechanism introduces the importance of strategic buyers for suppliers in modeling and decision making process. To illustrate the model efficiency, a prototype system has been modeled and is compared to the traditional tendering mechanism. The validation results confirm the model efficiency in providing long-term decisions in a dynamic environment.