基于时滞分割法的具有leakage项时滞的离散型神经网络状态估计

摘要：研究一类具有leakage时滞的离散型神经网络的状态估计问题．通过构造新的Lyapunov泛函得到保证估计误差全局渐近稳定的充分条件,并通过求解一个线性矩阵不等式（LMI）得到状态估计器的增益矩阵．采用一种新的时滞分割方法将变时滞区间分割为多个子区间,使该结果在获得更小的保守性同时也降低了计算的复杂度．     

潜艇运动对被动声纳浮标搜索的影响

由于被动声纳浮标隐蔽持续探测,潜艇可能由于机动进入浮标探测范围,从而增加了被探测概率。首先,在分析已有模型局限性的基础上,从潜艇运动对浮标搜索影响的原理出发,建立单枚浮标新增搜索区域模型,分析了单枚浮标搜索概率的影响因素,然后建模分析了潜艇运动对浮标拦截线搜索概率的影响因素,得出了影响作用仅对浮标作用距离、间距和航向敏感的结论,并对作战使用方法提出相关建议。     

基于飞蛾信息素寻偶机制的无人机集群协同搜索

为了提高无人机集群协同搜索移动目标的效率,提出一种基于飞蛾信息素寻偶机制的无人机集群协同搜索方法。根据飞蛾基于信息素选择飞行方向的寻偶行为,建立信息素图风向模型和飞蛾信息素寻偶模型。考虑无人机机间避撞约束,提出从飞蛾信息素寻偶机制到无人机集群分布式协同搜索的映射,并给出具体实现流程。仿真实验结果表明了所提方法在解决单个移动目标的协同搜索问题时的有效性和稳定性;外场飞行试验表明了所提方法在实际应用中的可行性。     

空中进攻作战出动计划禁忌搜索优化分析方法

空中进攻作战出动计划涉及多种复杂因素,大规模空中进攻作战出动计划不仅计算量大、出动强度高、架次多,而且当战场形势发生变化时,必须及时对飞机出动计划做出调整,采用手工方式制定和修改作战计划已难以适应未来空中进攻作战的要求。从分析制定空中进攻作战航空兵出动计划所涉及的诸要素出发,提出了一个较为合理的作战出动计划模型,并采用禁忌搜索求解模型的优良解,为实现计算机制定和及时修正空中进攻作战出动计划提出了一种有效的解决方法。     

卫星任务调度问题的约束规划模型

卫星任务规划与调度是空间资源管理的重要内容之一,其目的在于为卫星系统的任务计划编制提供科学合理的决策手段与依据。卫星任务调度问题的重要特点在于,调度任务存在可见时间窗口约束。只有在可见时间窗口内,调度任务才可能执行并完成。在进行合理假设的基础上,建立卫星任务调度问题的约束规划模型。对基本禁忌搜索算法进行改进,提出了模型求解的变邻域禁忌搜索算法。应用结果表明,约束规划模型的建立与求解是合理的。     

射频识别系统的多目标识别-反碰撞算法

对多目标识别技术的二进制搜索算法、时隙ALOHA算法进行了剖析,并对ALOHA算法用概率论和数理统计的方法进行了深入的理论分析,得出要根据所设计的射频识别系统的特点合理采用各种算法方可得到良好的效果.该算法的实现对提高高速公路上自动收费系统、人员和物流管理等各领域的工作效率具有一定的指导意义.     

What you should know about simulation and derivatives

Derivatives (or gradients) are important for both sensitivity analysis and optimization, and in simulation models, these can often be estimated efficiently using various methods other than brute‐force finite differences. This article briefly summarizes the main approaches and discusses areas in which the approaches can most fruitfully be applied: queueing, inventory, and finance. In finance, the focus is on derivatives of another sort. © 2008 Wiley Periodicals, Inc. Naval Research Logistics, 2008     

Column generation for stochastic green telecommunication network planning with switchable base stations

We present the green telecommunication network planning problem with switchable base stations, where the location and configuration of the base stations are optimized, while taking into account uncertainty and variability of demand. The problem is formulated as a two‐stage stochastic program under demand uncertainty with integers in both stages. Since solving the presented problem is computationally challenging, we develop the corresponding Dantzig‐Wolfe reformulation and propose a solution approach based on column generation. Comprehensive computational results are provided for instances of varying characteristics. The results show that the joint location and dynamic switching of base stations leads to significant savings in terms of energy cost. Up to 30% reduction in power consumption cost is achieved while still serving all users. In certain cases, allowing dynamic configurations leads to more installed base stations and higher user coverage, while having lower total energy consumption. The Dantzig‐Wolfe reformulation provides solutions with a tight LP‐gap eliminating the need for a full branch‐and‐price scheme. Furthermore, the proposed column generation solution approach is computationally efficient and outperforms CPLEX on the majority of the tested instances. © 2016 Wiley Periodicals, Inc. Naval Research Logistics 63: 351–366, 2016     

Investment in production resource flexibility: An empirical investigation of methods for planning under uncertainty

We examine several methods for evaluating resource acquisition decisions under uncertainty. Traditional methods may underestimate equipment benefit when part of this benefit comes from decision flexibility. We develop a new, practical method for resource planning under uncertainty, and show that this approach is more accurate than several commonly used methods. We successfully applied our approach to an investment problem faced by a major firm in the aviation information industry. Our recommendations were accepted and resulted in estimated annual savings in excess of $1 million (US). © 2003 Wiley Periodicals, Inc. Naval Research Logistics, 2003     

一类随机差分方程的解序列的最优停止规则

设报酬序列{x_(?),(?),n≥0}满足随机差分方程x_(n+1)=x_n+a_n+b_nε_(n+1)(ε_1,ε_2,…为白噪声序列)。本文讨论了用有限情形{x_n,0≤n≤N}的Snell包逼近无限情形{x_n,n≥0)的Snell包的条件,得到了x_n=E(x_n|(?))((?)=σ{ε_0,ε_1,…,ε_n},ε_0=0)的Snell包r_n的分解形式和最优停时存在的条件。最后讨论了最优停止规则的迭代计算法,并得出了迭代过程在有限步停止的充分条件。