D2TSAero Dataset for digital twin application in Aeronautics

The focus of this project is to develop a Digital Thread Twin architecture for Smart maintenance of Aeronautic systems. For the development of a reliable model that can ensure a reliable representation of the real system an Offline environment that mimics real system behavior was developed.

The simulation is based on the proposed model of an a Aircraft distribution System described in “Safety + ai: A novel approach to update safety models using artificial intelligence".

The proposed data set for the use case includes four Runs and five scenarios for each run. Five operating points for each run were simulated in order to represents healthy and Faulty conditions of the system. Faulty conditions are generated using MATLAB by injecting several faults categories at each epoch.

The proposed data concerns four components of the system mainly hydraulic pump,the three tanks, Engines, rear and front pumps. Dataset for 11 healthy and faulty scenarios was produced with several failure types for each of these components. Failures includes Noise NOI, Minor in Service Problem SER, Abnormal Instrument Reading AIR , External Leakage ELU and Parameter Deviation PDE. Noise on sensor was also considered to stimulate security threats.
Fault Type 1 : Noise on Instruments
Fault Type 2: Abnormal Instrument Reading
Fault Type 3: Minor in Service Problem
Fault Type 4: External Leakage
Fault Type 5: Parameter Deviation
Fault Type 6: Structural Deficiency
Fault Type 7: Breakdown

Features for Pumps : Pump flow for the five pumps
Features for Pump Hydraulic: Pump motor speed
Features for Driver: Driver Power
Features for Tank : Volume and temperature for the three tanks

The produced data set can be used to compare performance of different asset health estimation models and for the development of new predictive maintenance strategies.

This data empowers researchers to create new models for aeronautics systems maintenance management by leveraging digital twin technology. By using digital twins with muliagent systems and AI, scientists can predict and emulate the behavior of aircraft complex logic from these larger datasets.