Vulnerability Assessment of Intelligent Weapon Systems Based on Combined Weighting Method and Cloud Model

Abstract

With the rapid integration of artificial intelligence (AI) capabilities into weapon systems, vulnerabilities arising from the inherent limitations of AI technologies have emerged as a critical issue that jeopardizes system reliability and stability. This study presents a quantitative model for assessing the vulnerabilities of intelligent weapon systems by combining the weighting technique with cloud models. Through a review of literature and expert elicitation, a hierarchical index system comprising 5 primary indices and 14 secondary indices is constructed to facilitate vulnerability appraisal. The subjective Analytic Hierarchy Process is combined with the objective entropy method to calculate indicator weights for reducing subjectivity. The cloud model ingeniously transforms the ambiguity and randomness associated with evaluating qualitative indices into quantifiable cloud characteristics, thereby enabling the effective management of uncertainties. A case study assessing the vulnerability of a specific unmanned aerial vehicle system invites 10 field experts to validate the feasibility of the proposed methodology. The calculation of the similarity between the derived comprehensive cloud and predefined benchmark clouds can accurately classify the vulnerability level. The research conclusions provide essential guidelines for enhancing algorithm transparency, improving data diversity, and strengthening redundancy designs to mitigate vulnerabilities. With the accelerating integration of AI capabilities into weapon systems, it will be an ongoing imperative to continually assess and address vulnerabilities stemming from the inherent limitations of AI.