Abstract
Nowadays, unmanned aerial vehicles (UAV) become more applicable right along. The growing demand for their use is stipulated by economic considerations, and also by a capability for fulfilling the high-risk tasks. One of the most important tasks arising, when developing the unmanned equipment, is to detect dangerous situations because of possible failure of the on-board systems. Presently, this problem is solved mostly by multiple redundancies. Through computer technology development, along with traditional approaches, data mining tools, in particular artificial neural networks become more commonly used. The use of neural network tools to analyse multi-dimensional data can reduce the redundancy level, as well as to solve a wide range of tasks in real time. The paper suggests a new approach, which uses the multidimensional data analysis based on the neural network models, to develop an integrated system for assessing the UAV state. This system is designed to solve a wide range of tasks, such as troubleshooting the on-board equipment based on the complex analysis of measurements, redundancy of faulty sensors, assessment of the aerial vehicle state, and hazard prediction and prevention. Also, this system allows troubleshooting in the control system and enables the capability to complete a maneuver by assuming the control. Another important task is to keep logging of measurements and assess the aerial vehicle state using the neural network forecasting models. An equally important task is to verify the reliability of the UAV model comparing with real flight data. This system allows us not only to determine a divergence between the model and the object, but also localise the error and produce recommendations for correction. The paper presents a solution to these problems based both on the simulation results and on the real flight data.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call