Explainable AI framework for digital twin-empowered battlefield threat assessment

Artificial Intelligence (AI) is increasingly integrated into defense systems for real-time threat assessment and decision support. However, operational adoption remains limited due to concerns over transparency and trust. This study proposes an Explainable AI (XAI) framework for Digital Twin (DTW) -empowered battlefield threat assessment, combining multimodal deep learning and interpretable analytics. Heterogeneous data sources, including aerial imagery, acoustic signals, and sensor telemetry, are processed through unimodal modules ( Visual Deep and Acoustic Deep ) and fused via a DTW-based integration layer. Interpretability is achieved using SHAP (Shapley Additive exPlanations) for global and mission-specific insights. The experimental simulation results show that the proposed framework achieved 97.06% accuracy, 97.33% precision, 95.25% recall, and 97.29% specificity, outperforming state-of-the-art baselines. Moreover, SHAP analysis identified key threat indicators and sensor-level influences, enhancing situational awareness and operational trust.