Air Force Research Laboratory
Manuel D. Rossetti, David Boguslawski, Ryan Houx, Scott J. Mason, Mohsen Manesh, Ashlea Bennett, Joshua B. McGee
University of Arkansas
This research is intended to extend the knowledge base concerning logistical network modeling and design. Basic research techniques were developed to begin to model logistical networks within a hybrid simulation/analytic framework. The first step in this process is to develop robust approximations for portions of large-scale simulation models. This research examined the novel idea of utilizing neural networks as a meta-modeling technique to replace specific aspects of a simulation. This work started with the replacement of queuing stations within a logistics network. Any logistics network can be formulated as a network of material flowing through processes requiring resources. A new methodology was developed for forming approximations and improved approximators for queuing stations within a logistics network. The motivation for developing this approximation is the integration of such approximation in hybrid simulation/analytic methods for evaluating logistic networks. Future work should investigate the performance of these approximations within the larger logistical network context.
The use of information technology will allow for total asset visibility and the integration of logistical operations and logistical planning so that mission execution will be based on anticipatory and just-in-time strategies. Logistical planners must have the ability to rapidly evaluate the performance of various deployment scenarios in order to optimize the delivery of the correct materials, at the correct times, in the correct quantities to the correct locations. This research is intended to extend the knowledge base concerning logistical network modeling and design. Techniques will be developed to model military logistical networks within a hybrid simulation/analytic framework. The simulation/analytic framework will be used to efficiently predict the resulting performance of various logistical planning scenarios. In simulation/analytic approaches, the network is decomposed into subcomponents to which the most appropriate modeling technique is applied. The basic proposed approach is as follows. Given the network topology, demand characterization, and resource configuration, the method: 1) randomly generates a snapshot of demand at a instant in time, 2) simulates the routing and resource scheduling allocation, and loading assignment for each of the demands generated, 3) loads or rejects each demand based on the routing and scheduling analysis, 4) updates the state of the network prior to evaluating each demand, 5) exhausts the list of potential demands in the snapshot, 6) model the performance of state of the network at the network, delivery, and resource levels, 7) summarize the performance measures over multiple snapshots, and 8) reports statistics from the modeled network. Computational efficiency is gained since only portions of the logistical network are simulated; approximations are used to estimate other performance metrics.