Operations Analysis Using Bayesian Networks

Goh Sing Huat

Abstract

This project is concerned with the management of uncertainty in military planning and explored into the Human Fix approach to manage uncertainty using Bayesian networks.

To demonstrate the applicability of Bayesian networks in military planning, two Bayesian network models were fictitiously created - one is concern with National Defence at strategic level and the other is concern with Unit Operations at tactical level. It was shown how the models could be updated with the latest information to evaluate the battle situation, so as to facilitate the military planner in his planning. With the formulation of the network model, sensitivity analysis was done to examine the relationships between the event nodes and the end node. This will help to enable scarce military resources to be allocated effectively to determine the state of nodes critical to the end node.

Numerical and structural changes where also made to the models to evaluate different battle situations and scenarios. Though the two models illustrated are static in nature, they do illustrate the the dynamism of battle evaluation under ever changing situations through them.