【作者】 万梅芳；

【导师】 陈以；

【作者基本信息】 桂林电子科技大学 ， 控制理论与控制工程， 2009， 硕士

【摘要】 物流,因其贯穿了生产和流通的全过程,所以,合理、高效的物流能够协调与完善整个生产和流通结构,产生巨大的利润,成为企业和国民经济新的利润增长点。我国无论是国民经济领域还是企业经济领域,物流成本都居高不下,在宏观层次上表现为物流成本占GDP的比重过高,在微观层次上表现为物流成本位居企业成本项目的首位。我国物流科学研究起步较晚,由于有效的物流成本控制所产生的经济效益和社会效益都是显著的,因此,物流成本控制如何实现合理化或最优化是本论文关注和研究的课题。最经济控制(the Most Economical Control,简称MEC)是以系统的技术性能指标为约束条件,经济效益为目标函数的最优控制。最经济控制与智能控制相结合产生了最经济智能控制(the Most Economical Intelligent Control,简称MEIC)。神经网络是智能控制中常用的一类,由于采用梯度下降法的BP(Back Propagation)神经网络算法存在易产生局部极小而产生伪收敛,甚至不收敛的缺点,本文采用径向基函数(Radial Based Function,简称RBF)神经网络对物流系统实现最经济智能控制研究。论文在系统分析径向基函数神经网络和遗传算法(Genetic Algorithm,简称GA)理论的基础上,对最经济控制的研究方法进行了深入的研究,建立了神经网络的网络代价函数,给出了一种基于改进的自适应遗传算法的径向基函数神经网络的结构优化算法,实现了神经网络的最经济控制。采用自适应的交叉算子和变异算子,不仅增强了样本的多样性,而且扩大了搜索空间。由于构造物流系统的经济代价函数比较复杂,文中对最经济控制的代价函数进行了分解,将物流代价函数转换为物流系统的成本函数,建立了物流系统的订货量和成本之间的函数,得到了物流系统的经济代价函数,并将最经济控制算法应用于物流系统的研究中,实现了物流系统的最经济控制,对物流成本的研究具有一定的参考价值。更多还原

【Abstract】 Logistics runs through the whole process of production and circulation. The ra-tional and high-e?cient logistics can coordinate and perfect the production structureand entire circulation, produce the enormous profits. It could become the new pointof growth profit for enterprises and national economy. The logistics costs of our coun-try reached high levels not only in the field of national economy but also enterpriseeconomic. To macroscopically level it is shown as too high logistics cost proportion ofto GDP while logistics cost occupy enterprise cost the first place of project at microlevel. The logistics scientific research is started relatively late in our country. Becauseit is remarkable both in economic benefit and social benefit caused by the e?ectivecontrolling of logistics cost, how to realize rationalized or optimal goal for logistics costcontrol become the focus in this paper.The most economical control (MEC) is the optimal control constrained by tech-nical performances and with the economical cost as its object function. The mosteconomical intelligent control (MEIC) combines MEC with intelligent control. Neuralnetwork control is a common intelligent control. Based on gradient decline algorithm,it is easy to fall into the local minimum value, which will result in false convergence.Therefore, in this paper radical based function neural network (RBFNN) has been usedto the research on the most economical control in logistics system.Based on the analysis of RBFNN and GA, the most economical control is de-scribed in detail. The network cost function of RBFNN is created, and a structural op-timization algorithm based on the improved genetic algorithm for RBF neural networkis deserved. In this paper, adaptive crossover probability and mutation probability,which not only increase the diversity of samples, but also expand the search space, areused to improve genetic algorithm. The most economical control of neural network isimplemented by using the improved algorithm and network cost. As the complexityof the logistical economical cost function, the cost function of the MEC is analyzed,and the logistics economical cost function is changed into the logistical cost function.An equation between the ordering quantity and the cost of the logistics system is got,and the economical cost function of the logistics system is also deserved. Lastly, theeconomical control in logistics system is got by using the economical control algorithm,which has some reference value to the logistics cost research.更多还原