模糊生产系统中的多类型差异作业平行机批调度优化

针对现实生产制造系统中存在的时间参数模糊化问题,采用梯形模糊数表征时间参数,给出了一种具有模糊加工时间、模糊交货期与模糊批次间隔的,以最小化制造跨度和最小化提前/拖期惩罚为目标的多类型差异作业平行机批调度问题模型。在问题求解方面,给出了一种具有量子行为的,采用混沌局部优化的混合粒子群算法,避免求解过程陷入局部最优。仿真实验验证了该算法具有可行性和有效性。     

模糊环境下的差异作业平行机批调度优化问题研究

针对现实生产制造系统中存在的时间参数模糊化问题,采用梯形模糊数表征时间参数,给出了一种具有模糊加工时间与模糊批次间隔的,以最小化制造跨度为目标的差异作业平行机批调度问题模型。在问题求解方面,给出了一种基于粒子群优化和差异进化的混合优化算法,避免求解过程陷入局部最优,并通过改进的Batch First Fit算法获得优化的分批。仿真实验验证了该算法具有可行性和有效性。     

多点收缩混沌优化方法及全局收敛性证明

针对目前混沌优化算法在选取局部搜索空间时的盲目性,提出一种具有自适应调节局部搜索空间能力的多点收缩混沌优化方法.该方法在当前搜索空间搜索时保留多个较好搜索点,之后利用这些点来确定之后的局部搜索空间,以达到对不同的函数和当前搜索空间内已进行搜索次数的自适应效果.给出了该算法以概率1收敛的证明.仿真结果表明该算法有效的提高了混沌优化算法的性能,改善了混沌算法的实用性.     

具有模糊目标的多目标双矩阵对策集结与求解

针对不确定性多冲突环境,建立了多个具有模糊目标的多目标双矩阵对策的综合集结模型.在假定局中人各模糊目标的隶属函数为线性函数的情形下,基于总体模糊目标的可达度,给出了纳什均衡解的定义,并应用粒子群优化算法对集结模型求解.最后,给出一个军事例子说明了模型的实用有效性和粒子群优化算法求解的高效性.     

求解柔性作业车间调度的动态群智能优化算法

针对柔性作业车间调度问题,提出一种新型两阶段动态混合群智能优化算法.算法初始阶段采用动态邻域的协同粒子群进行粗搜索,第二阶段提出了基于混沌算子的蜂群进行细搜索,既增强了种群多样性,又提高了算法搜索精度,实现了全局搜索与局部搜索能力的有效平衡.针对柔性作业车间调度问题特点,采用独特的编码方式和位置更新策略来避免不合法解的产生.最后将此算法在不同规模的实例上进行了仿真测试,并与最近提出的其他几种具有代表性的算法进行了比较,验证了算法的有效性和优越性.     

粒子群优化模糊神经网络在语音识别中的应用

针对模糊神经网络训练采用BP算法比较依赖于网络的初始条件,训练时间较长,容易陷入局部极值的缺点,利用粒子群优化算法(PSO)的全局搜索性能,将PSO用于模糊神经网络的训练过程.由于基本PSO算法存在一定的早熟收敛问题,引入一种自适应动态改变惯性因子的PSO算法,使算法具有较强的全局搜索能力.将此算法训练的模糊神经网络应用于语音识别中,结果表明,与BP算法相比,粒子群优化的模糊神经网络具有较高的收敛速度和识别率.     

混沌粒子群神经网络在上证指数预测中的应用

上证指数预测是一个非常复杂的非线性问题,为了提高对上证指数预测的准确性,本文采用基于混沌粒子群(CPSO)算法对BP神经网络算法改进的方法来进行预测.BP神经网络算法目前已经应用到预测、聚类、分类等许多领域,取得了不少的成果.但自身也有明显的缺点,比如易陷入局部极小值、收敛速度慢等.用混沌粒子群算法改进BP神经网络算法的基本思想是用混沌粒子群算法优化BP神经网络算法的权值和阈值,在粒子群算法中加入混沌元素,提高粒子群算法的全局搜索能力.对上证指数预测的结果表明改进后的预测方法,具有更好的准确性.     

基于混合单纯形算法的模糊均值图像分割

针对模糊C均值算法用于图像分割时对初始值敏感、容易陷入局部极值的问题,提出基于混合单纯形算法的模糊均值图像分割算法.算法利用Nelder-Mead单纯形算法计算量小、搜索速度快和粒子群算法自适应能力强、具有较好的全局搜索能力的特点,将混合单纯形算法的结果作为模糊C均值算法的输入,并将其用于图像分割.实验结果表明:基于混合单纯形算法的模糊均值图像分割算法在改善图像分割质量的同时,提高了算法的运行速度.     

基于混沌优化和自适应反向学习的秃鹰搜索算法

针对秃鹰搜索算法求解精度低、收敛速度较慢、容易陷入局部最优的问题,提出一种基于混沌优化和自适应反向学习的秃鹰搜索算法.首先,在选择搜索空间阶段引入正弦混沌映射更新秃鹰群位置,增加随机性,优化全局搜索能力;其次,在俯冲捕获猎物阶段加入指数自适应,平衡了全局搜索和局部搜索,同时加快收敛速度;最后,对更新后的最优秃鹰位置使用反向学习策略,提高跳出局部最优的可能性.选取12个测试函数对算法的性能进行了测试,结果表明本文改进的秃鹰搜索算法具有更优的求解精度和收敛速度.     

基于直觉模糊软集的群决策方法研究

研究了专家权重未知情况下基于直觉模糊软集的群决策问题.首先利用距离提出了一种基于直觉模糊软集的决策算法.然后通过考虑专家个体提供信息本身的不确定性和专家之间提供的信息间的一致性,定义了直觉模糊软集的知识测度和基于α-相似关系的一致度,由此提出了一种确定专家权重的方法.进而给出了一种基于直觉模糊软集的群决策算法.最后通过实例说明所提出算法是有效的与合理的.     

带时间窗偏好的同时配集货且需求可拆分车辆路径问题

针对带时间窗偏好的同时配集货且需求可拆分车辆路径问题,最小化派遣成本、理货成本、时间窗惩罚成本以及油耗成本之和,建立数学模型。设计混合遗传变邻域搜索算法求解问题,在算法中引入时空距离的理念,首先用最近邻插入法和Logistic映射方程生成初始种群;然后利用变邻域搜索算法的深度搜索能力优化算法;提出自适应搜索策略,平衡种群进化所需的广度和深度;设计拆分准则,为各客户设置不同的拆分服务量;提出确定车辆最优出发时间的时差推移法,减少车辆在客户处的等待时间;最后通过多组算例验证本文模型和算法的有效性。     

重力输水系统

为学习微积分的读者介绍如何将基本流体力学应用于对发展中国家非常重要的简单重力输水系统的建模。首先推导出Bernoulli方程,从而了解作为沿流线运动的流体质点压力、速度和高度之间的关系。其次,应用Bernoulli方程分析一个简单的输水系统的合力和流速。然后,对层流和湍流分别考虑分压水箱、不同直径的管道及摩擦的影响。最后,讨论在密克罗尼西亚和洪都拉斯重力输水系统的设计和安装。     

Discontinuous Local Delivery Costs Re-Examined

The problem of locating depots when discontinuities occur in the local delivery cost function has been examined before and a heuristic procedure proposed. Here, an alternative search procedure, which can be used in conjunction with the previous heuristic, is described leading to the development of a family of heuristics. Some of these are tested and the results given. Conclusions are then drawn as to the best procedure to use under different circumstances.     

Network Design for Express Shipment Delivery

Service network design problems arise at airlines (passenger and cargo), trucking companies, railroads, etc., wherever there is a need to determine cost minimizing routes and schedules, given constraints on resource availability and level of service. We focus on a particular service network design application, namely, the express shipment delivery problem, and we develop models and a solution technique designed specifically for large-scale express delivery problems with time windows. Using data from an express delivery company, we demonstrate that our approach can produce savings in total operating costs and provide a valuable tool for making decisions at strategic and tactical levels.     

Delivery performance in vendor selection decisions

Increased reliance upon outsourcing has made the issue of vendor selection even more critical to the success of the modern manufacturing organization. The usual performance measure on which selection is based has been the distribution of the vendor’s delivery lead-time (LT), often as characterized by the mean and variance. In this paper, we show that the distribution of demand per unit time (DPUT) must also be considered if an optimal decision is to be made.     

配送量不定的供应链协调

本文研究供应链管理中配送量不确定时，制造商和分销商之间的协调问题。通过建立惩罚和奖励机制模型来协调供应链各部分的运作，使得分散系统就如一个整体。同时分析供应链的利润分配问题，分别得出各个机制下分销商的最优决策。     

Delivery delay variation in multi-echelon inventory problems

Mathematical models for controlling multi-echelon inventory and production levels have been developed for entire interconnected systems rather than myopically for each point in the inventory or production system. With such models come the additional burden of considering the lead times and delays associated with the movement of items through the network. The delays develop when a backorder state occurs and are often considered in models by approximating the mean delay by an application of Little's Law and without consideration of the delay variation. In this article a general methodology for incorporating the variation of the delay is derived and incorporated into a backorders optimisation model. The result of incorporating delay variation into the model is a tolerance interval for order quantities that may be utilised for tighter inventory control.     

Batch Delivery Scheduling on a Single Machine

The problem of partitioning a set of independent and simultaneously available jobs into batches and sequencing them for processing on a single machine is presented. Jobs in the same batch are to be delivered together, upon completion of the last job in the batch. Jobs finished before this time have to wait until delivery. There are a delivery cost depending on the number of batches formed and an earliness cost for jobs finished before delivery. The dynamic programming approach to minimizing the total cost is considered, yielding two pseudopolynomial algorithms when the number of batches has a fixed upper bound. A polynomial algorithm for a special case of the problem is also presented.     

The Delivery Man Problem with time windows

In this paper, a variant of the Traveling Salesman Problem with Time Windows is considered, which consists in minimizing the sum of travel durations between a depot and several customer locations. Two mixed integer linear programming formulations are presented for this problem: a classical arc flow model and a sequential assignment model. Several polyhedral results are provided for the second formulation, in the special case arising when there is a closed time window only at the depot, while open time windows are considered at all other locations. Exact and heuristic algorithms are also proposed for the problem. Computational results show that medium size instances can be solved exactly with both models, while the heuristic provides good quality solutions for medium to large size instances.