基于进化规划的自适应高斯神经网络

自适应高斯神经网络能够对目标信号的功率谱有效识别特征进行自动提取和分类,但此网络使用BP算法,其误差能量函数是一个不规则的超曲面,容易陷入局部极小值.因此,提出了一种使用进化规则来设计和训练自适应高斯神经网络的新方法.该方法能够自动地确定网络的最优结构和联结权值,同时避免网络的局部优化.实验结果表明,将该方法用于被动声纳目标的分类识别,能够有效地克服局部最小问题,具有更好的识别率.     

精确保障的理论研究与发展

根据未来信息化战争的新特点,探讨了适应信息化战争保障需求、基于信息网络技术、资源重构技术、柔性保障机制和系统理论的精确保障的内涵,研究归纳出当今美军装备保障的特点、提出精确保障的理论研究应该充分借鉴系统理论、网络中心战理论、敏捷制造理论的先进思想,指出了我军精确保障实现的途径以及应该遵循的原则.     

A decomposition algorithm applied to planning the interdiction of stochastic networks

We describe the application of a decomposition based solution method to a class of network interdiction problems. The problem of maximizing the probability of sufficient disruption of the flow of information or goods in a network whose characteristics are not certain is shown to be solved effectively by applying a scenario decomposition method developed by Riis and Schultz [Comput Optim Appl 24 (2003), 267–287]. Computational results demonstrate the effectiveness of the algorithm and design decisions that result in speed improvements. © 2005 Wiley Periodicals, Inc. Naval Research Logistics, 2005.     

汽油调合优化模型

为了提高炼油厂制订成品油调合方案的科学性,研究了汽油调合优化的非线性规划模型,给出了目标函数和约束条件的具体形式。根据模型特征,选用模拟退火算法对模型求解。最后,通过某炼油厂的一个应用实例验证了上述成品油调合优化模型的有效性。     

基于多智能体的反导传感器任务规划算法

首先探讨了反导作战中各类传感器资源的构成特点和功能划分,然后基于多智能体multy-agent system(MAS)的分布式人工智能技术提出了适应反导作战需求的传感器任务规划体系架构,既而深入分析了体系架构中各类agent的主要功能,最后给出了一个实体层传感器的任务规划流程和算法实例,为构建高效率、智能化的反导作战多传感器任务规划技术提供了一种有效的研究方法。     

Partial equilibrium in risk‐based production decisions

Characteristically, a small subset of operational problems admit risk neutrality when contingent claims methodology were used in their analysis. That is, for the majority of manufacturing and production problems, operating cash flows are not directly linked to prices of traded assets. However, to the extent that correlations can be estimated, the methodology's applicability to a broader set of operational problems is supported. Our article addresses this issue with the objective of extending the use of contingent claims techniques to a larger set of operational problems. In broad terms, this objective entails a partial equilibrium approach to the problem of valuing uncertain cash flows. To this end, we assume risk aversion and cast our approach within Merton's intertemporal capital asset pricing model. In this context, we formulate a “generic” production valuation model that is framed as an exercise in stochastic optimal control. The model is versatile in its characterization and can easily be adapted to accommodate a wide‐ranging set of risk‐based operational problems where the underlying sources of uncertainty are not traded. To obtain results, the model is recast as a stochastic dynamic program to be solved numerically. The article addresses a number of fundamental issues in the analysis risk based decision problems in operations. First, in the approach provided, decisions are analyzed under a properly defined risk structure. Second, the process of analysis leads to suitably adjusted probability distributions through which, appropriately discounted expectations are derived. Third, through consolidating existing concepts into a standard and adaptable framework, we extend the applicability of contingent claims methodology to a broader set of operational problems. The approach is advantageous as it obviates the need for exogenously specifying utility functions or discount rates.© 2011 Wiley Periodicals, Inc. Naval Research Logistics, 2011     

Optimal path finding in direction,location, and time dependent environments

This article examines optimal path finding problems where cost function and constraints are direction, location, and time dependent. Recent advancements in sensor and data‐processing technology facilitate the collection of detailed real‐time information about the environment surrounding a ground vehicle, an airplane, or a naval vessel. We present a navigation model that makes use of such information. We relax a number of assumptions from existing literature on path‐finding problems and create an accurate, yet tractable, model suitable for implementation for a large class of problems. We present a dynamic programming model which integrates our earlier results for direction‐dependent, time and space homogeneous environment, and consequently, improves its accuracy, efficiency, and run‐time. The proposed path finding model also addresses limited information about the surrounding environment, control‐feasibility of the considered paths, such as sharpest feasible turns a vehicle can make, and computational demands of a time‐dependent environment. To demonstrate the applicability and performance of our path‐finding algorithm, computational experiments for a short‐range ship routing in dynamic wave‐field problem are presented. © 2012 Wiley Periodicals, Inc. Naval Research Logistics, 2012     

直接序列扩频信号仿真数据的高效生成方法

扩频技术在通信和卫星导航领域得到了广泛应用,传统仿真数据的生成方法存在计算量大、忽略信道影响因素,较难实现高精度生成等问题.结合扩频系统信号和噪声的特点,提出了基于扩频码码片相关的信号生成以及基于AR模型的带限高斯噪声生成两种高效仿真信号生成方式.该方法计算量相对于传统高精度方法可由1GHz的生成速率降低至码片速率级的数据生成速率,同时保证了信号生成的灵活性和可控性.相关结果可用于现代化导航接收机的仿真数据快速生成.     

Models and algorithms for shuffling problems in steel plants

In this article, we study item shuffling (IS) problems arising in the logistics system of steel production. An IS problem here is to optimize shuffling operations needed in retrieving a sequence of steel items from a warehouse served by a crane. There are two types of such problems, plate shuffling problems (PSP) and coil shuffling problems (CSP), considering the item shapes. The PSP is modeled as a container storage location assignment problem. For CSP, a novel linear integer programming model is formulated considering the practical stacking and shuffling features. Several valid inequalities are constructed to accelerate the solving of the models. Some properties of optimal solutions of PSP and CSP are also derived. Because of the strong NP‐hardness of the problems, we consider some special cases of them and propose polynomial time algorithms to obtain optimal solutions for these cases. A greedy heuristic is proposed to solve the general problems and its worst‐case performances on both PSP and CSP are analyzed. A tabu search (TS) method with a tabu list of variable length is proposed to further improve the heuristic solutions. Without considering the crane traveling distance, we then construct a rolling variable horizon heuristic for the problems. Numerical experiments show that the proposed heuristic algorithms and the TS method are effective. © 2012 Wiley Periodicals, Inc. Naval Research Logistics, 2012     

Economic lot sizing with constant capacities and concave inventory costs

This article studies the classical single‐item economic lot‐sizing problem with constant capacities, fixed‐plus‐linear order costs, and concave inventory costs, where backlogging is allowed. We propose an O(T³) optimal algorithm for the problem, which improves upon the O(T⁴) running time of the famous algorithm developed by Florian and Klein (Manage Sci18 (1971) 12–20). Instead of using the standard dynamic programming approach by predetermining the minimal cost for every possible subplan, we develop a backward dynamic programming algorithm to obtain a more efficient implementation. © 2012 Wiley Periodicals, Inc. Naval Research Logistics, 2012