Multiobjective path finding in stochastic dynamic networks, with application to routing hazardous materials shipments

Multiobjective path finding in stochastic dynamic networks, with application to routing hazardous materials shipments

0.00 Avg rating0 Votes
Article ID: iaor20062109
Country: United States
Volume: 39
Issue: 3
Start Page Number: 383
End Page Number: 399
Publication Date: Aug 2005
Journal: Transportation Science
Authors: , ,
Keywords: programming: probabilistic, programming: multiple criteria
Abstract:

We describe a method for finding nondominated paths for multiple routing objectives in networks where the routing attributes are uncertain, and the probability distributions that describe those attributes vary by time of day. This problem is particularly important in routing and scheduling of shipments of very hazardous materials. Our method extends and integrates the work of several previous authors, resulting in a new algorithm that propagates means and variances of the uncertain attributes along paths and compares partial paths that arrive at a given node within a user-specified time window. The comparison uses an approximate stochastic dominance criterion. We illustrate the effects of changing primary parameters of the algorithm using a small test network, and we show how the nondominated solution set achieved is larger than the set that would be identified if the uncertainty in routing attributes were ignored. We then demonstrate how the algorithm creates an effective solution set in a case study using a large network.

Reviews

Required fields are marked *. Your email address will not be published.