Modern industrial production is only possible due to continuous planning. Due to the complexity of production systems, domain experts planning future production rely heavily on software support to provide a data basis for sound decisions. These decision support systems need not only to provide information but also to provide it on time, as time spent waiting for the support system delays decisions, making production sluggish and inflexible. Therefore, quick response times of support systems are a significant competitive advantage. An example from the semiconductor industry is utilization thresholds, allowing a planner to quickly validate a production plan concerning its expected adherence to cycle time targets. These thresholds are significantly affected by the material flow characteristics of the equipment in use. With the help of simulation, they can be determined easily, but only after a significant amount of time. In this work, we demonstrate how equipment-specific utilization thresholds can be provided nearly instantly with the help of data farming and machine learning, creating a surrogate model. This approach effectively moves the calculation heavy and, therefore, time-consuming simulation runtime into the project development phase, allowing for near-instantaneous response times during application. We discuss all components of this approach, from feature selection to simulation on a large scale as required for data farming to the use of available automated machine learning approaches to train neural networks, which in turn are capable of providing decision support almost instantly. Therefore, the resulting fast response system is capable of providing planners with timely decision support, allowing them to better utilize resources, thus improving the balance between cycle time and production cost.     «

Modern industrial production is only possible due to continuous planning. Due to the complexity of production systems, domain experts planning future production rely heavily on software support to provide a data basis for sound decisions. These decision support systems need not only to provide information but also to provide it on time, as time spent waiting for the support system delays decisions, making production sluggish and inflexible. Therefore, quick response times of support systems are...     »