Data-Driven and AI-Integrated Virtual Simulation Experimental Teaching Reform for Aircraft Systems

Abstract

To address the challenges in virtual simulation experimental teaching for aircraft systems—such as stringent airworthiness constraints, high costs, and insufficient process-oriented evaluation—this paper proposes an experimental teaching framework that integrates
data-driven methodologies with artificial intelligence. Leveraging flight data, a high-fidelity aircraft system simulation model is established.
Machine learning and knowledge graph techniques are incorporated to enable intelligent fault diagnosis, personalized learning support, and
optimized process evaluation. A four-dimensional progressive teaching model is designed to form a competency-oriented closed-loop cultivation system. Taking the virtual simulation experiment for aircraft avionics system fault diagnosis as a case study, a national-level first-class
virtual simulation experimental course has been developed and implemented across multiple universities. Results indicate that this approach
enhances the realism and interactive intelligence of experimental teaching, while significantly strengthening students' engineering practice and
problem-solving abilities.