A hybrid decision making model for evaluating land combat vehicle system

Abstract

New Land Combat Vehicles (LCV) are being acquired to improve the fighting capability of the Australian Defence Force (ADF). Combat vehicle design has traditionally focussed on the use of new and improved technologies to provide an appropriate level of survivability and lethality against a range of threats. However, other factors such as sensors, firepower, mobility, tactics, situational awareness and communications also play a role in increasing or decreasing the survivability and lethality of LCV. For the most part, LCV system interactions between the components are defined by logical or physical links which complicate the evaluation of its capabilities. In addition, the different types of operational scenarios and threats provide a challenge to the Decision Maker (DM) in providing a clear answer to the problem of determining the ‘best’ combination of LCV system configurations. This paper utilises two methodologies, Bayesian Network (BN) and Multi-Criteria Decision Analysis (MCDA), and presents a hybrid decision making model for evaluating the operational impact of different LCV system specifications and configurations. BN is employed to establish a qualitative and quantitative representation of the relations between the variables of the model and calculate standard values of uncertain LCV capabilities like survivability, lethality, etc. MCDA is adopted to integrate the influence of LCV capabilities and calculate the utility value of the selected options. In order to analyse a very large number of model outputs from all possible system components, a prototype tool of the hybrid approach has been developed. This tool conveniently performs inferences over the BN, examines various proposed configurations in a more intuitive and friendly fashion using data analysis, visualisation and sensitivity analysis techniques, and also searches for the ‘best’ configuration for the LCV system. The tool utilises public-release BN software (GeNIe), statistical packages (R) and a built-in Analytic Hierarchy Process (AHP) to allow the required calculations to be completed automatically and the results captured in both tabular and graphical form.