A simulation-based and flexi-time applying prediction model for scheduling personnel deployment times in the production planning process of cyber-physical systems

在工业4.0的背景下，预期的生产计划和控制的分散化给人员规划带来了新的挑战。即使人类仍然是智能工厂的关键因素，人工活动也会发生变化。部署时间将发生变化，工作内容将需要更具体的资格，从而降低部署的连续性。同时，由于机器和订单的自主性与实时控制相结合，很难提前推断出具体的需求时间；相反，它将在实际需求期间被识别。目前还没有办法应对这些变化了的情况。为了确保有效的资源规划，尽管由此产生的时间冲突，该项目正在开发一种基于模拟的预测模型，用于分散生产系统中操作人员的需求。通过包括灵活的工作时间，人员能力以这样一种方式适应需求，即分散控制系统可以考虑到秩序、机器和最重要的人员的总体目标。对于后者，不仅追求公司的目标，也追求员工个人的目标。基于仿真的方法考虑了现实中发生的随机不确定性，例如波动的过程时间或故障。预测模型侧重于短期规划范围。在第一步中，使用控制参数（例如人员部署时间）生成场景，并根据专注于目标函数的优化器进行更改。高解决方案质量的场景被组合成一个随机的总体计划，并由此导出确定性的部署时间。此外，还审查了该方法的可行性、先决条件、潜力和限制（例如工作时间模型的影响）。

The expected decentralization of production planning and control within the context of Industry 4.0 poses new challenges for personnel planning. Even if the human being remains a key factor in the smart factory, manual activities will change. Deployment times will shift and work content will require more specific qualifications, resulting in a reduced deployment continuity. At the same time, due to the autonomy of machines and orders in combination with real-time control, it is hardly possible to deduce the concrete times of requirement in advance; instead, it will be just identified during the actual requirement. There are currently no methods available for coping with these changed conditions. In order to ensure efficient resource planning despite the resulting time conflict, the project is developing a simulation-based forecasting model for operative personnel requirements in decentralized production systems. By including flexible working times, the personnel capacity is adapted to the demand in such a way that the decentralized control system can take into account the overall objectives of order, machine and above all personnel. For the latter, not only company goals but also goals of individual employees are pursued.The simulation-based approach takes into account stochastic uncertainties that occur in reality, such as fluctuating process times or failures. The forecast model focuses on a short-term planning horizon. In a first step, scenarios are generated using control parameters (e.g. personnel deployment times) and varied according to an optimizer focusing on the target function. Scenarios of high solution quality are combined to a stochastic overall plan, from which deterministic deployment times are derived. Furthermore, the methodology is examined for feasibility, prerequisites, potentials and limits (e.g. influence of working time models).