Research On The Workflow Pattern Mining Technology

Workflow management systems which are support process control are widely used in order to meet the needs of the high efficient automatic production. Process mining can use historical data, such as the event logs to generated abstract process model, and provide favorable conditions for the deployment of workflow system. One of the important tasks of process mining is to describe the relationship between events in the process, also known as process discovery.In this paper, firstly through extensive investigation and research, present domestic and foreign research situation of process mining field, understand the general process of process mining, combing the different categories of process mining algorithm, find that the traditional process mining algorithms have defects in the ability of mining process structure and anti-noise, and some methods based on intelligent computing take too much time. In order to overcome these problems, based on Petri net and Workflow net, extended the concept of causal matrix used in the paper to representation process model. Then presents a general process mining framework based on heuristic optimization algorithm, then apply the tabu search algorithm to process mining task. Some key problems, such as program initialization, creation of tabu list and neighborhood are discussed in detail. Then in order to achieve good balance between global optimization and local optimization to obtain higher algorithm efficiency, apply the modified bat algorithm in process mining area, demonstrates the realization and parameter selection of bat algorithm. At last implement two algorithm as plug-ins of ProM. The experiment verified the correctness of the process mining framework, both two process mining approach can deal with different flow structures, robustness to noise and less time consuming.