A Python Extension to Simulate Petri nets in Process Mining

Abstract

The capability of process mining techniques in providing extensive knowledge and insights into business processes has been widely acknowledged. Process mining techniques support discovering process models as well as analyzing process performance and bottlenecks in the past executions of processes. However, process mining tends to be "backward-looking" rather than "forward-looking" techniques like simulation. For example, process improvement also requires "what-if" analyses. In this paper, we present a Python library which uses an event log to directly generate a simulated event log, with additional options for end-users to specify duration of activities and the arrival rate. Since the generated simulation model is supported by historical data (event data) and it is based on the Discrete Event Simulation (DES) technique, the generated event data is similar to the behavior of the real process.

SynBPS: a parametric simulation framework for the generation of event-log data

“…Compared with the existing solutions based on BPMN models, 33,34,37 Petri nets, 35,39,40 and process trees 36 found in the literature review, this framework uses Markov chains to represent the control flow of a simulation model. This approach has the disadvantage that it does not support concurrent/parallel processing of events, unless a significantly higher number of states is introduced (with multiple activities processed in parallel encoded as separate states).…”

“…Petri Nets Simulator (PNSIM) 35 extend the ability to generate a simulation model from observed behavior in an event log to performing changes to distributions of the activity durations and arrival rates. The control flow is represented by a Petri net, and arrival times and activity durations are represented by distributions (which are not specified further).…”