Neural Network Enhancement of the Los Alamos Force Deployment Estimator
This document has been relocated to http://scholarsmine.mst.edu/ele_comeng_facwork/1766
There was 1 download as of 28 Jun 2016.
Abstract
The Force Deployment Estimator (FDE) is a decision support system. It allocates transportation resources given inputs such as forces to be deployed and their desired arrival times. Other inputs are assumptions about conditions that affect performance: carrier start time, node capacity, sustainment shipping time, bulk sustainment per day, ammo sustainment per day, unit start time, carrier service time, carrier round trip time, and carrier reassignment time. Outputs include the mean and standard deviation of estimated unit arrival times versus goal times, and data files for post-processing. However, when a goal time is not met, the simulator gives no explanation of why. This is difficult to do because of the volume of data involved. Poor allocation choices are buried in a mountain of other decisions, whose affects are difficult to assess individually.
To find the most troublesome allocations, we separate the cases that give the worst results. A neural network identifies the decisions that are common to these. We apply a similar procedure to the cases where outputs are good. We report as suspect the decisions that occur only in the former cases. The neural network for this system needs to be capable of clustering data with no apriori knowledge of correct output categories. It also needs to be able to handle inexact (fuzzy) determinations of these categories. Finally, it needs to be able to handle large data patterns without large sets of example cases. We have chosen Adaptive Resonance Theory with fuzzy input/output representation, which fits all these criteria.
