USING MIXTURES of TRUNCATED EXPONENTIALS for SOLVING STOCHASTIC PERT NETWORKS

In this paper, we transform a PERT network into a mixtures of truncated exponentials Bayesian network. We use the Shenoy-Shafer architecture to propagate the MTE potentials in the resulting MTE PERT Bayes net and thus to find the marginal distribution of the project completion time. Finding the distribution of the project completion time is important because there is no closed form expression for the distribution of the maximum of two normal distributions and this fact, previously forced the researchers to make false assumptions about its distribution. In this research, we show that by approximating the maximum of two distributions using MTE’s a very accurate estimation for the project completion time can be obtained.