MODELLING AND CLASSIFICATION OF AIRCRAFT SENSOR AND ACTUATOR FAILURES USING NEURAL NETWORKS

Abstract

The classification of aircraft sensor and actuator faults is crucial for ensuring flight safety, system reliability, and operational efficiency. This article presents the modeling and classification of aircraft sensor and actuator failures using a neural network-based approach. The proposed model utilizes a fourth-order Runge-Kutta method combined with a Non-linear Auto-Regressive Network with exogenous inputs to detect and classify failures effectively. The model's performance is evaluated through Matlab simulations, testing various parameter sets, including R(ω_x, ω_y), R(ω_z, ω_y), and R(ω_x, ω_z) over a fixed duration of t = 50 s. The simulation results demonstrate the model's capability in accurately identifying and classifying failures, providing critical support for fault-tolerant adaptive control systems in aircraft. By leveraging neural network techniques, this method enhances real-time failure detection and classification, contributing to improved system reliability and safety. The simulation results validate the potential of integrating machine learning into aviation diagnostics, paving the way for more robust and adaptive aircraft control systems in the future.