Self-optimizing decentralized production control

面对多变的市场、日益多样化的产品以及随之而来的日益复杂的制造系统，生产控制在制造业中尤为重要。由于需要更频繁地调整工艺链和生产资源，集中式生产控制方法正在达到极限。相比之下，分散控制方法的特点是高度的灵活性和快速的适应性。随着网络物理生产系统的发展，现在首次出现了能够实现与订单有关的全面数据采集、生产组件之间的通信和使用本地计算能力的技术。这是采用分散控制方法的必要先决条件。然而，从方法论的角度来看，算法的局部优化倾向是另一个障碍。基于经验的考虑全局最优的局部decisionmaking.The拟议项目的主要目标是分散的生产控制方法，允许自主和可变的决策与同时基于经验的自我优化方面的生产系统的全局系统性能的基础研究，可以消除这个缺点。该方法应特别考虑车间生产的要求。为了实现项目的目标，一个合适的系统架构，分散生产控制的CPPS的框架内，首先检查。随后，设想一种方法，允许分散和可变的决策与经验为基础的全球系统性能的考虑。为此所需的估值基础是由一系列有待审查的关键数字构成的。为后续的研究搭建了测试环境。为此，首先创建了一个具有代表性的车间生产仿真模型。控制模拟模型内物料流的软件实现的控制方法与此相关。最后，在结构化测试计划的框架内进行了模拟实验，该计划提供了正在研究的自主学习控制方法的潜力和限制的深入和科学的大量知识。

Facing a volatile market, an increasing variety of products and consequently increasingly complex manufacturing systems, production control is of particular importance in manufacturing. Due to the need for more frequent adaptation of process chains and production resources, centralized production control approaches are reaching their limits. In contrast, decentralized control approaches are characterized by high flexibility and quick adaptability. With the development of cyber-physical production systems (CPPS), technologies are now available for the first time with which comprehensive order-related data acquisition, communication between production components and the use of local computing capacities can be realized. This is an essential prerequisite for the introduction of decentralized control approaches. From a methodological point of view, however, local optimization tendencies of the algorithms represent a further obstacle. Experience-based consideration of the global optimum in local decision-making can eliminate this disadvantage.The main objective of the proposed project is the fundamental research of a method for decentralized production control that allows autonomous and variable decision-making with simultaneous experience-based self-optimization with regard to the global system performance of the production system. The method should take special account of the requirements of shop-floor production. In order to achieve the project goals, a suitable system architecture for decentralized production control is first examined within the framework of a CPPS. Subsequently, a method is conceived that allows decentralized and variable decision making with experience-based consideration of global system performance. The valuation basis required for this is formed by a system of key figures to be examined. A test environment is set up for the subsequent research. For this purpose, a simulation model of representative shop-floor production is first created. The software-implemented control method that controls the material flow within the simulation model is linked to this. Finally, simulation experiments are carried out within the framework of a structured test plan, which provide in-depth and scientifically substantial knowledge of the potentials and limits of the autonomous and learning control method being researched.