BP人工神经网络的应用及其实现技术

简述了BP人工神经网络的算法原理 ,利用Batchnet程序对一个三层BP人工神经网络进行了设计 ,应用Matlab中人工神经网络工具箱设计了一个用于函数逼近的BP人工神经网络 ,最后 ,运用Matlab中可视化工具Simulink对一个BP人工神经网络例子进行了仿真 ,通过例子 ,探讨了BP人工神经网络的应用并介绍目前几种实用的BP人工神经网络实现技术 .     

An efficient heuristic for the multi-item newsboy problem with a single constraint

This article treats the problem of determining optimal and approximately optimal order quantities for a multiple-item inventory system subject to a single constraint on space or budget. Although this problem can be solved by the usual method of Lagrange multipliers, we wish to consider a more efficient scheme that requires fewer computations. We provide calculations that compare and contrast four approximation techniques. In particular, we have discovered a method that yields a direct algebraic expression of the problem parameters for allocation and achieves an expected profit within 90% of the optimal in about 90% of the cases tested.     

Minimizing sums and products of linear fractional functions over a polytope

In this paper, we develop efficient deterministic algorithms for globally minimizing the sum and the product of several linear fractional functions over a polytope. We will show that an elaborate implementation of an outer approximation algorithm applied to the master problem generated by a parametric transformation of the objective function serves as an efficient method for calculating global minima of these nonconvex minimization problems if the number of linear fractional terms in the objective function is less than four or five. It will be shown that the Charnes–Cooper transformation plays an essential role in solving these problems. Also a simple bounding technique using linear multiplicative programming techniques has remarkable effects on structured problems. © 1999 John Wiley & Sons, Inc. Naval Research Logistics 46: 583–596, 1999     

Properties of an approximate hazard transform

The calculation of the exact reliability of complex systems is a difficult and tedious task. Consequently simple approximating techniques have great practical value. The hazard transform of a system is an invertible transformation of its reliability function which is convenient and useful in both applied and theoretical reliability work. A simple calculus for finding an approximate hazard transform for systems formed by series and parallel combinations of components is extended so that it can be used for any coherent system. The extended calculus is shown to lead to conservative approximations. A first order version of the extended calculus is also discussed. This method of approximation is even more simple to use, but is not always conservative. Examples of its application indicate that it is capable of giving quite accurate results.     

Approximation results for min‐max path cover problems in vehicle routing

This article studies a min‐max path cover problem, which is to determine a set of paths for k capacitated vehicles to service all the customers in a given weighted graph so that the largest path cost is minimized. The problem has wide applications in vehicle routing, especially when the minimization of the latest service completion time is a critical performance measure. We have analyzed four typical variants of this problem, where the vehicles have either unlimited or limited capacities, and they start from either a given depot or any depot of a given depot set. We have developed approximation algorithms for these four variants, which achieve approximation ratios of max{3 ‐ 2/k,2}, 5, max{5 ‐ 2/k,4}, and 7, respectively. We have also analyzed the approximation hardness of these variants by showing that, unless P = NP , it is impossible for them to achieve approximation ratios less than 4/3, 3/2, 3/2, and 2, respectively. We have further extended the techniques and results developed for this problem to other min‐max vehicle routing problems.© 2010 Wiley Periodicals, Inc. Naval Research Logistics, 2010     

Computation techniques for large scale undiscounted markov decision processes

In this paper we consider computation techniques associated with the optimization of large scale Markov decision processes. Markov decision processes and the successive approximation procedure of White are described. Then a procedure for scaling continuous time and renewal processes so that they are amenable to the White procedure is discussed. The effect of the scale factor value on the convergence rate of the procedure and insights into proper scale factor selection are given.     

Normal and weibull distributions

Numerous applications of the Weibull distribution in diverse fields of human endeavor are well known today. Nevertheless, it is not uncommon to find applications of the normal distribution in such fields of studies as agriculture, biology, chemistry, engineering, physics, sociology and others. At the present time we have at our disposal many more refined statistical techniques for analyzing the normal rather than the Weibull data. Consequently, it is important for applied statisticians to know if some of their data which can be described by the Weibull distribution can also be described by the normal distribution. The present investigation of the author reveals that the normal distri,bution can be considered to be a good approximation to the Weibull distribution as long as its shape parameter is in the open interval (3.25, 3.61). This fact enables them to perform a refined statistical analysis of their data. Conversely, they can now easily compute the desired normal cumulative probabilities from the Weibull distribution function, which would be especially helpful for those standard normal deviates whose cumulative probabilities cannot be read from the available tables of normal cumulative probability. In a similar situation they can also use the Weibull distribution to obtain an approximation to any desired normal deviate for a given normal probability which may be better than those obtained by the linear interpolation method.     

Design of follow‐up experiments for improving model discrimination and parameter estimation

One goal of experimentation is to identify which design parameters most significantly influence the mean performance of a system. Another goal is to obtain good parameter estimates for a response model that quantifies how the mean performance depends on influential parameters. Most experimental design techniques focus on one goal at a time. This paper proposes a new entropy‐based design criterion for follow‐up experiments that jointly identifies the important parameters and reduces the variance of parameter estimates. We simplify computations for the normal linear model by identifying an approximation that leads to a closed form solution. The criterion is applied to an example from the experimental design literature, to a known model and to a critical care facility simulation experiment. © 2004 Wiley Periodicals, Inc. Naval Research Logistics, 2004     

平面 NURBS 曲线的椭圆弧自适应逼近

给出了用椭圆弧及双椭圆弧自适应逼近平面NURBS曲线的算法。算法所得到的椭圆样条能够G1连续 ,双椭圆样条还能够保形。与现行的圆弧逼近算法相比 ,本算法不需要求解非线性方程组 ,而是由给定的插补误差自动计算参数增量 ,得到椭圆曲线的特征点 ,还可以将误差控制在预期的范围之内 ;与现行的直线插补方法相比 ,不需要额外的时间和空间 ,也适用于CNC环境。本算法在腔体加工、二维轮廓加工等方面有特别的实用价值。     

神经网络的函数逼近理论

分析了将函数逼近理论与方法引入神经网络研究的必要性；从经典函数逼近与统计分析两方面详细地讨论了多层前馈网（ＭＬＰ）逼近能力分析的基本方法及结论；分析了正则理论观点下的径向基函数网络（ＲＢＦ）的逼近能力；讨论了ＲＢＦ网与多层前馈网在最佳逼近特性上的差异。文末指出了神经网络函数逼近的发展方向。     

Sensitivity of queueing models to the assumption of exponentiality

This study deals with determining the effects on queuing models of assuming that interarrival and service times are exponentially distributed. Steady-state expected value measures of congestion are compared for a variety of GI/G models when assuming an M/M model as an approximation. The majority of this effort involves single-channel queues; however, some multichannel situations are also considered. In addition, effects of the exponential approximation are observed in certain cost-optimization models.     

Stein–Chen approximation and error bounds for order fill rates in assemble‐to‐order systems

Assemble‐to‐order (ATO) is an important operational strategy for manufacturing firms to achieve quick response to customer orders while keeping low finished good inventories. This strategy has been successfully used not only by manufacturers (e.g., Dell, IBM) but also by retailers (e.g., Amazon.com). The evaluation of order‐based performance is known to be an important but difficult task, and the existing literature has been mainly focused on stochastic comparison to obtain performance bounds. In this article, we develop an extremely simple Stein–Chen approximation as well as its error‐bound for order‐based fill rate for a multiproduct multicomponent ATO system with random leadtimes to replenish components. This approximation gives an expression for order‐based fill rate in terms of component‐based fill rates. The approximation has the property that the higher the component replenishment leadtime variability, the smaller the error bound. The result allows an operations manager to analyze the improvement in order‐based fill rates when the base‐stock level for any component changes. Numerical studies demonstrate that the approximation performs well, especially when the demand processes of different components are highly correlated; when the components have high base‐stock levels; or when the component replenishment leadtimes have high variability. © 2012 Wiley Periodicals, Inc. Naval Research Logistics, 2012     

Decomposition-based approximation algorithms for the one-warehouse multi-retailer problem with concave batch order costs

We study the one-warehouse multi-retailer problem under deterministic dynamic demand and concave batch order costs, where order batches have an identical capacity and the order cost function for each facility is concave within the batch. Under appropriate assumptions on holding cost structure, we obtain lower bounds via a decomposition that splits the two-echelon problem into single-facility subproblems, then propose approximation algorithms by judiciously recombining the subproblem solutions. For piecewise linear concave batch order costs with a constant number of slopes we obtain a constant-factor approximation, while for general concave batch costs we propose an approximation within a logarithmic factor of optimality. We also extend some results to subadditive order and/or holding costs.     

属性约简与近似集的关系研究

上近似和下近似是在粗集理论中最基本的概念,而粗集理论的主要思想又是在保持分类能力不变的前提下通过对属性的约简导出问题的决策和分类规则。本文讨论了属性的约简和相对约简与上近似和下近似的关系,并在此基础上提出了一种新的相对约简———保近似约简。     

A mathematical programming approach to identification and optimization of a class of unknown systems

There exists a class of decision problems for which: (1) models of input-output response functions are not available in a closed-form, functional representation; (2) informational costs associated with learning about the response function are significant. For these problems, combining identification with optimization using mathematical programming is potentially attractive. Three approaches to the identification-optimization problem are proposed: an outer-linearized approximation using relaxation (OLR); an inner-linearized approximation using restriction (ILR); and a sequential combination of inner- and outer-linearized subproblems (SIO). Algorithms based on each approach are developed and computational experience reported.     

跳跃-滑翔弹道扰动引力自适应网格快速赋值方法

为实现高超声速跳跃-滑翔弹道扰动引力的快速赋值,提出自适应网格赋值模型,并根据反距离加权理论,优化广义延拓逼近算法,对模型的逼近误差进行分析。该赋值模型的网格划分为两级,第一级网格根据标准弹道空域进行划分,第二级网格根据滑翔导弹实际弹道在线生成。根据一级网格节点数据,通过优化广义延拓逼近算法计算二级网格节点数据,最后根据二级单元内插计算实际弹道点的扰动引力值。仿真结果表明:在同等大小的网格划分下,优化广义延拓自适应网格模型的逼近精度高于一般赋值方法;在同等精度要求下,该赋值模型的最大单元格边长大于一般赋值方法,从而减少了单元格划分数量,进而降低弹上数据存储量;针对不同滑翔方向以及不同滑翔距离的跳跃-滑翔弹道,该模型逼近误差对应的落点偏差小于5 m,具有较好的适应性。该赋值模型在满足计算速度的前提下,提高了传统赋值方法的逼近精度,降低了弹上存储量,具有一定的工程应用价值。     

Pricing bridges to cross a river

We consider a pricing problem in directed, uncapacitated networks. Tariffs must be defined by an operator, the leader, for a subset of m arcs, the tariff arcs. Costs of all other arcs in the network are assumed to be given. There are n clients, the followers, and after the tariffs have been determined, the clients route their demands independent of each other on paths with minimal total cost. The problem is to find tariffs that maximize the operator's revenue. Motivated by applications in telecommunication networks, we consider a restricted version of this problem, assuming that each client utilizes at most one of the operator's tariff arcs. The problem is equivalent to pricing bridges that clients can use in order to cross a river. We prove that this problem is APX‐hard. Moreover, we analyze the effect of uniform pricing, proving that it yields both an m approximation and a (1 + lnD)‐approximation. Here, D is upper bounded by the total demand of all clients. In addition, we consider the problem under the additional restriction that the operator must not reject any of the clients. We prove that this problem does not admit approximation algorithms with any reasonable performance guarantee, unless P = NP, and we prove the existence of an n‐approximation algorithm. © 2007 Wiley Periodicals, Inc. Naval Research Logistics, 2007     

Distribution network design: Selection and sizing of congested connections

This paper focuses on certain types of distribution networks in which commodity flows must go through connections that are subject to congestion. Connections serve as transshipment and/or switching points and are modeled as M/G/1 queues. The goal is to select connections, assign flows to the connections, and size their capacities, simultaneously. The capacities are controlled by both the mean and the variability of service time at each connection. We formulate this problem as a mixed integer nonlinear optimization problem for both the fixed and variable service rate cases. For the fixed service rate case, we prove that the objective function is convex and then develop an outer approximation algorithm. For the variable service rate case, both mean and second moment of service time are decision variables. We establish that the utilization rates at the homogeneous connections are identical for an optimal solution. Based on this key finding, we develop a Lagrangian relaxation algorithm. Numerical experiments are conducted to verify the quality of the solution techniques proposed. The essential contribution of this work is the explicit modeling of connection capacity (through the mean and the variability of service time) using a queueing framework. © 2005 Wiley Periodicals, Inc. Naval Research Logistics, 2005.     

Continuous-time markov decision processes with nonzero terminal reward

In this article we consider a continuous-time Markov decision process with a denumerable state space and nonzero terminal rewards. We first establish the necessary and sufficient optimality condition without any restriction on the cost functions. The necessary condition is derived through the Pontryagin maximum principle and the sufficient condition, by the inherent structure of the problem. We introduce a dynamic programming approximation algorithm for the finite-horizon problem. As the time between discrete points decreases, the optimal policy of the discretized problem converges to that of the continuous-time problem in the sense of weak convergence. For the infinite-horizon problem, a successive approximation method is introduced as an alternative to a policy iteration method.     

Stochastic regret minimization for revenue management problems with nonstationary demands

We study an admission control model in revenue management with nonstationary and correlated demands over a finite discrete time horizon. The arrival probabilities are updated by current available information, that is, past customer arrivals and some other exogenous information. We develop a regret‐based framework, which measures the difference in revenue between a clairvoyant optimal policy that has access to all realizations of randomness a priori and a given feasible policy which does not have access to this future information. This regret minimization framework better spells out the trade‐offs of each accept/reject decision. We proceed using the lens of approximation algorithms to devise a conceptually simple regret‐parity policy. We show the proposed policy achieves 2‐approximation of the optimal policy in terms of total regret for a two‐class problem, and then extend our results to a multiclass problem with a fairness constraint. Our goal in this article is to make progress toward understanding the marriage between stochastic regret minimization and approximation algorithms in the realm of revenue management and dynamic resource allocation. © 2016 Wiley Periodicals, Inc. Naval Research Logistics 63: 433–448, 2016