Toward Explainable Decision Support in Aerospace Design: Integrating Human Expertise and Computational Insights

This paper addresses the development of explainable decision support systems (DSS) for aerospace design through the integration of human expertise and computational intelligence. Traditional machine learning methods and high-fidelity simulation models often lack transparency, which constrains engineer trust and complicates human–system interaction in safety-critical design contexts. To overcome these limitations, a combined methodological framework is considered that unifies explainable artificial intelligence (XAI), human-in-the-loop decision-making, physics-informed neural networks (PINNs), and digital twins within a coherent design support paradigm. Explainable models enable interpretable assessment of the influence of design parameters on optimization outcomes, while human-in-the-loop mechanisms ensure continuous expert validation and alignment with safety regulations and operational standards. The use of PINNs guarantees physical consistency of computational models by embedding governing physical laws into data-driven representations. Digital twins further provide an interactive platform that integrates simulation results, measurement data, and expert judgments for comparative analysis of alternative design scenarios. The results demonstrate that such hybrid and explainable frameworks enhance transparency, reproducibility, and reliability of aerospace design decisions, while facilitating effective collaboration between engineers and intelligent systems. The study identifies key perspectives for future research, including deeper integration of physical modeling into explainable algorithms, advancement of visualization techniques, and the development of synergistic platforms combining digital twins and human-in-the-loop approaches for multi-objective optimization under uncertainty.