The order-driven single and small series batch production is characterized by a high variety and the demand for short delivery times. Therefore, a robust production planning is essential. Usually the master data, which is used for scheduling, does not represent the current conditions on the shop floor, due to its static nature. Thus the planning results are of poor quality. To counteract that, this article describes an approach that uses production-condition-dependent execution times of all throughput time components under the consideration of uncertainty. The provision of the scheduling parameters is done with the help of a knowledge-based system that uses a previously on actual data set up case and rule base. A simulation based validation showed that this approach led to an adherence to delivery dates of 93 % in comparison to the utilization of static master data with 74 %.