Collaborative evolutionary multi-project resource scheduling

Abstract

Large, real-world resource constrained multiple-project scheduling problems involve complex quality criteria. Clearly, issues such as meeting resource constraints, and minimising slippage, flow-time and tardiness are all appropriate and will be applied in such cases. However, often crucial criteria exist which are difficult or impossible to formalise. Since such large projects span multiple resources and often long time-spans, these extra criteria may include seasonal issues, deep knowledge of the scheduler concerning resource issues at different times, or perhaps knowledge and experience concerning the individual project managers involved in the projects. We justify and describe a collaborative evolutionary scheduling system, which enables the search to be guided by both the standard schedule quality criteria and also the master scheduler's non-formalised knowledge and experience. Preliminary experiments are described which indicate great promise for the approach.