Quantification of central input factors for the determination of planning lead times in shop floor production

在生产结构复杂的公司中，计划订单交货期的确定仍然是一个未解决的问题。因此，公司经常在按时交付给客户方面遇到问题。除了由于工作流程延迟而导致对时间表的遵守不足之外，过早的工作流程还会导致可避免的库存。需要精确的计划订单交货时间，以便向客户确认可行的截止日期，计划生产能力并更好地组织采购。然而，在实践中，经常使用严格的方法来确定计划的订单交货时间，这对不断变化的环境影响没有充分的反应。可能的环境影响包括短期请病假或生产负荷的波动。确定计划订单交货期的经典方法特别包括基于估计、历史值、逻辑模型和模拟的方法。这些方法考虑了不同的信息和数据进行测定。大多数经典方法的局限性主要表现在假设过于简化，难以获得高质量的规划。模拟等方法抵消了这些限制，但会导致非常高的应用程序工作量。特别是在复杂的生产环境中，如车间生产（如工具和专用机器制造），由于工作内容分散，每个作业的操作次数波动，在实践中无法实现精确的计划。此外，对于合同制造商来说，计划交货时间的确定不足直接影响到对客户的遵守进度。在车间生产中，影响订单交货期的主要输入因素尚未得到系统的研究。因此，本项目的目的是从六个合同制造商的公司数据中确定和量化车间生产计划订单交货时间的中心输入因素。该项目将使用机器学习（ML）来识别数据中的模式并推导广义假设。对中心因果关系的洞察将帮助被调查公司更准确地预测计划订单的交货时间，从而提高他们满足客户最后期限的能力。

The determination of planned order lead times is still an unsolved issue in companies with complex production structures. As a result, companies often have problems delivering to customers on time. In addition to insufficient adherence to schedules due to late work processes, premature work processes lead to avoidable inventory. Precise planned order lead times are required to confirm a feasible deadline to the customer, to plan production capacities and to better organize procurement.In practice, however, rigid methods for determining the planned order lead times are often used, which do not react sufficiently to changing environmental influences. Possible environmental influences are e.g. short term sick calls or fluctuations of the load of production. Classical approaches for the determination of planned order lead times include, in particular, methods based on estimates, historical values, logistic models and simulation. These methods take into account different information and data for the determination. The limitations of most of the classical methods are especially in too simplified assumptions, which makes a high planning quality difficult. Approaches such as simulation counteract these limitations, but lead to a very high application effort. Especially in complex production environments like shop floor production (e.g. tool and special machine manufacturing), precise planning is not implemented in practice due to scattering work content and a fluctuating number of operations per job. In addition, for contract manufacturers, insufficient determination of the planned lead times has a direct impact on the adherence to schedules to customers. The central input factors influencing the order lead times in a shop floor production are not yet systematically investigated.Therefore, the aim of this project is to identify and quantify central input factors for the determination of planned order lead times in shop floor production from company data of six contract manufacturers. Machine learning (ML) will be used in the project to identify patterns in data and to derive generalized assumptions. The insights into the central cause-effect relationships will help the investigated companies to more precisely forecast planned order lead times and thus improve their ability to meet their customers' deadlines.