Limited lookahead control policies adapt the concept of supervisor synthesis in discrete-event systems (DES) to work for very large and/or dynamic systems. This is accomplished by considering an N-step projection of possible behaviours of the system, instead of dealing directly with the entire system. This paper builds on previous work to produce a limited lookahead algorithm that incorporates probabilistic event occurrences and a desirability function for strings of events. An example is used to compare the results from this algorithm with those obtained using Markov chain analysis. Results are shown to be consistent, and future application of the algorithm to more complex problems is discussed.

No abstract