Multi-Fidelity Digital Twin Framework for General Aviation Fault Diagnosis
A novel method for intelligent fault diagnosis in general aviation aircraft has been introduced, combining multi-fidelity digital twin technology with enhancements from FMEA knowledge. This framework tackles issues such as limited real fault data, a variety of fault types, and weak fault signatures. It consists of four components: high-fidelity flight dynamics simulation utilizing the JSBSim six-degree-of-freedom (6-DoF) engine, FMEA-based fault injection modeling for 19 engine fault types, multi-fidelity residual feature extraction using paired-mirror and GRU surrogate prediction residuals, and interpretable report generation enhanced by a large language model (LLM). The digital twin produces 23-channel engine health monitoring data through semi-empirical sensor synthesis equations. Details of this method can be found in a paper on arXiv with the identifier 2604.22777.
Key facts
- The method uses multi-fidelity digital twin and FMEA knowledge enhancement.
- It integrates four modules: flight dynamics simulation, FMEA-driven fault injection, multi-fidelity residual extraction, and LLM-enhanced report generation.
- JSBSim six-degree-of-freedom (6-DoF) engine is used for high-fidelity simulation.
- 23-channel engine health monitoring data is generated via semi-empirical sensor synthesis equations.
- A three-layer fault injection engine models 19 engine fault types based on FMEA.
- Multi-fidelity residual computation includes paired-mirror residuals and GRU surrogate prediction residuals.
- The framework addresses scarce real fault data, diverse fault types, and weak fault signatures.
- The paper is published on arXiv with ID 2604.22777.
Entities
Institutions
- arXiv