Applying intelligent data analysis technologies for detecting damages to UAVs
I. K. Romanova
Abstract
In this work, features of Data Mining (DM) have been utilized in order to ensure their active applications in the problem of damage detection. Approaches for identifying damages in the process of UAV movement have been proposed. Multiple simulations of normal and damaged vehicles movements, which underwent both a change in the shape of the body and distortion of the sensor readings due to icing, have been carried out. Calculations have been carried out for both deterministic and random characteristics of damage to the elements of the investigated object. Both independent and cross combinations of both object shape and sensors have been considered. About 30 status indicators have been tested and the most reliable indicators of damage have been identified for both shape changes and sensor distortions. Damage is identified in terms of indicator calculations which are based on the results of measurements of angular velocities, trajectory inclination and angles of attack. The ability to assess the presence of this damage has been shown in this work.