Determination of routings and process time information from event logs

Abstract: Production planning and scheduling decisions are based on underlying models and their parameters which are typically considered granted, accurate and up-to-date. However, changes and uncertainties may force the production to change, shift and adapt in its settings, consequently, the models and parameters used in the planning and execution phases lose their validity. Usually, these changes are not explicitly known, but are incorporated the manufacturing execution system data. To cope with the missing, uncertain or wrong data, process and data mining should be repeatedly employed on the available manufacturing event logs. The paper describes an analysis framework which is able to extract the required information from large-scale and noisy event logs and offers the ability to infer and restore a more realistic history of the underlying manufacturing process hidden by random noise and consequent errors. 1