【作者】 剡昌锋；

【导师】 商向东； 郑鹏；

【作者基本信息】 沈阳工业大学 ， 机械制造及其自动化， 2002， 硕士

【摘要】 自动化仓库是集物料搬运、计算机控制、管理和仓储等科学与技术的一门综合性学科，由于它具有节约劳动力，作业迅速准确，可以提高保管效率，降低物流费用等优越性，日益受到人们越来越多的重视。它是工厂物流、柔性制造系统和计算机集成制造系统中不可或缺的重要组成部分，同时也在现代物流系统和交通运输领域得到越来越广泛的应用。要保证整个仓库系统的性能指标最优，就要对各子系统进行调度，使其协调运转，也就是自动化仓库的调度问题。自动化仓库的调度已经成为国内外学者研究的一个很重要的课题。 本文首先对目前国内外自动化仓库的现状和研究调度问题的方法进行了全面地综述。通过对邮包自动化仓库系统中的自动导引小车运输问题和巷道堆垛机的拣选顺序认真地分析，建立起了调度的基本数学模型，提出了用总的空载时间作为目标函数，并且把这两类问题都归结为求解旅行商问题。提出了邮包自动化仓库在线调度的概念和方法，即动态调度问题，实际上等同于解决在线旅行商问题。在此基础上，作者提出了用改进遗传算法来解决这一问题，通过用MatLab语言编程，分别对改进遗传算法、3-opt的启发式方法和先进先出的调度方法进行计算机模拟仿真，最后对这三种方法进行了全面地比较，实验结果清楚地表明可以用改进遗传算法这种方法来解决邮包自动化仓库中的两类调度问题。同时具有很高的稳定性、向最优解收敛的速度更快和能极大地提高调度效率的能力，将来应用于工程实际的价值很高。 另外，本文还针对目前国内在对堆垛机控制方面普遍采用的三段式减速控制规律的弊病，提出用两个RBF神经网络分别作为位置误差学习器和点位位置控制器，由于神经网络的自学习功能，可以使得在控制当中不断校正和修改在训练时由于样本不足造成的缺憾。通过对两个RBF神经网络的训练和仿真，证明这种速度控制规律具有自适应性强、响应快、定位准确和无超差等优点。更多还原

【Abstract】 Automated warehouse is complex field that integrates material handling, computer controlling, management and so on. It is paid more and more attention, because it is of saving labor force, decreasing logistics costs. It is necessary part in factory logistics, flexible manufacture material and computer integral manufacture system, and applied in modern material flow system, traffic and transport wider and wider. If the performance of whole automated warehouse is ensured optimum, the sub-systems must be scheduled and run very well, that is, the scheduling problems of automated warehouse. Moreover it has become a very important research subjection studied by researcher in china and abroad.In the paper, the situation and method of scheduling problem studied of automated warehouse are surveyed completely. The problems of automate guide vehicle transport and stack crane picking order are analyzed deeply in parcel post automated warehouse, basic mathematic model of scheduling is set up. The conception and method of online scheduling in parcel post automated warehouse are put forward, that is dynamic scheduling problem, which is equal to solve online traveling salesman problem. On the basis of mentioned above, an improving genetic algorithms to solve this problem is put forward. Improving genetic algorithms, 3-opt heuristic and first input first output is simulated separately in computer by MatLab, and compared completely. The result of experiment shows that improving genetic algorithms can be used to obtain solution of two kinds of scheduling problems in parcel post automated warehouse, stability is strong, and the speed converging to optimum is higher. This result to improve the efficiency of scheduling, and will benefit practical engineering.Furthermore, because the three grades decelerate control method, which is used widely, in stack crane control in china has some shortcoming, two RBF neuron network applied in position error study and dot position control is put forward. Because neuron network is of self-study, emendation and modification can be achieved in control and can remedy the shortcoming caused by shortage of sample. The result of simulation experiment shows that the speed control rule is of strong adaptable, high-speed response, accuracy position and no speed overshoot.更多还原