Efficient search of Petri Nets for deadlock-free scheduling in FMSs using heuristic functions

This paper presents and evaluates the performance of three heuristic functions, based on Petri Nets (PN), which are used to optimize the average flow time in flexible manufacturing systems. The developed heuristic functions aim to reduce the complexity of the scheduling problem by searching only the necessary portion of the Petri Net reachability, graph. In addition, each heuristic function is equipped with a parameter to provide a trade-off between the solution quality and the search effort. A major advantage of these heuristics is in the case of modification to deal with other measures of performance such as resources utilization and due date measures. An experimental study was performed using these heuristic functions on randomly generated test cases. A new routine for automatically generating PN models from a production plan is presented. The Average Operation Waiting Time (AOWT) heuristics function is found to outperform the other two functions, Remaining Processing Time (RPT) and Scheduling with Dispatching Rules (SDR), with respect to the obtained average flow time and the solution CPU time. In addition, the newly developed heuristics help in scheduling larger problems with high efficiency compared with the results reported in the literature.