基于APS的离散制造业生产预测的研究与应用

发布时间：2017-12-30 17:45

本文关键词：基于APS的离散制造业生产预测的研究与应用　出处：《宁夏大学》2016年硕士论文　论文类型：学位论文

【摘要】：为了实现企业的动态管理,企业采用高级计划排程技术,它是一种优化计划,可将时间、订单、库存等生产中真实存在的问题考虑进去。在企业生产中随时获取各种动态变化,来调整生产去迎合市场。帮助企业解决不能动态平衡实时物料需求和资源的问题,为离散制造业提供了有效支撑。而在APS中用数学模型预测零件生产数量为企业生产计划提供参照具有巨大的实用意义。但企业的生产预测模型却具有复杂性、多层次性、开放性、非静态性等特点,因此对它建立模型十分困难。本文基于某离散制造企业的背景,该企业是一家制造阀门的企业,实现的是多品种、少批量、多批次的生产方式,使企业对环境变化具有更大的柔性。在满足客户需求的情况下实现企业最大利益下的最小成本目标,就必须采用更有效地生产组织方式,合理地利用库存,为了完善企业自身已有的APS系统,企业需要建立离散型生产线的生产预测作为发展的主要方向。通过访问企业的数据库获取制造阀零件的历史数据,发现企业生产线上的数据具有不规律、有噪声及变量之间存在的耦合等复杂特点,首先对原始数据预处理,经过了数据清洗、数据集成以及归一化。运用BP神经网络、RBF神经网络算法建立生产预测模型,利用遗传算法的全局搜索以及收敛速度快的特征对模型优化处理,并对仿真结果进行分析比对,最后选择了一种最优的算法模型应用在企业中,在APS的实际排程中发挥了重大的作用。
[Abstract]:In order to realize the dynamic management of enterprises, enterprises with advanced planning and scheduling technology, it is a kind of optimization plan, the time order, the real problems in the production of the inventory account. To get all kinds of changes in enterprise production, to adjust production to meet the market. Can help enterprises solve the dynamic balance of real-time material needs and resources, to provide effective support for the discrete manufacturing industry. And the APS used in the mathematical model to predict the number of parts production has great practical significance to provide reference for enterprise production planning. But the production of pre enterprise measurement model has a complex, multi-level, open, non static, so on it is very difficult to establish the model. In this paper, a discrete manufacturing enterprise based on the background, the company is a manufacturer of valve enterprises, to achieve multi varieties, small batch, multi batch production mode, the The enterprise is more flexible to the change of the environment. To achieve the minimum cost maximum interests of enterprises under the circumstances to meet customer demand, we must adopt more effective production organization, reasonable use of inventory, in order to improve the existing APS system of enterprises, enterprises need to establish a discrete production line as the main production forecast the direction of development. Access to the history data of manufacturing valve parts by accessing the enterprise database, find enterprise production line data is not the rule, there exists between noise and variable coupling characteristics, firstly, the pre-processing of the original data, after data cleaning, data integration and normalization. Using BP neural network, RBF neural network the algorithm to establish production prediction model, using genetic algorithm's global search and fast convergence characteristics of the optimization model, and the simulation results are analyzed. At the end of this paper, an optimal algorithm model is applied to the enterprise and plays a significant role in the actual scheduling of APS.

【学位授予单位】：宁夏大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：F426.4;F273;TP183

【相似文献】