A simulation environment for design and testing of aircraft adaptive fault‐tolerant control systems

Abstract

PurposeThe purpose of this paper is to present the development of a Matlab/Simulink‐based simulation environment for the design and testing of indirect and direct adaptive flight control laws with fault tolerant capabilities to deal with the occurrence of actuator and sensor failures.Design/methodology/approachThe simulation environment features a modular architecture and a detailed graphical user interface for simulation scenario set‐up. Indirect adaptive flight control laws are implemented based on an optimal control design and frequency domain‐based online parameter estimation. Direct adaptive flight control laws consist of non‐linear dynamic inversion performed at a reference nominal flight condition augmented with artificial neural networks (NNs) to compensate for inversion errors and abnormal flight conditions following the occurrence of actuator or sensor failures. Failure detection, identification, and accommodation schemes relying on neural estimators are developed and implemented.FindingsThe simulation environment provides a valuable platform for the evaluation and validation of fault‐tolerant flight control laws.Research limitations/implicationsThe modularity of the simulation package allows rapid reconfiguration of control laws, aircraft model, and detection schemes. This flexibility allows the investigation of various design issues such as: the selection of control laws architecture (including the type of the neural augmentation), the tuning of NN parameters, the selection of parameter identification techniques, the effects of anti‐control saturation techniques, the selection and the tuning of the control allocation scheme, as well as the selection and tuning of the failure detection and identification schemes.Originality/valueThe novelty of this research efforts resides in the development and the integration of a comprehensive simulation environment allowing a very detailed validation of a number of control laws for the purpose of verifying the performance of actuator and sensor failure detection, identification, and accommodation schemes.

Evolutionary Algorithm for Artificial-Immune-System-Based Failure-Detector Generation and Optimization

“…Comparisons between detection rates for optimized and unoptimized schemes are presented. This second set of results is based upon data collected from the WVU F-15 flight simulation [34], performed using the WVU Motus six-degree-of-freedom flight simulator. The air vehicle model implemented emulates the NASA Intelligent Flight Control System F-15 research aircraft.…”

“…The lookup tables are divided to represent the aerodynamic contributions from individual control surfaces (including canards and dual rudder) to accommodate the modeling of actuator failures. Direct adaptive fault-tolerant control laws [34] are included, generating commands for the differential canard, collective and differential stabilator, differential aileron, and rudder.…”

“…In this example, a three-dimensional self is considered, consisting of neural-network outputs for roll, pitch, and yaw rates (NNp, NN q , and NN r ). The neural networks are part of a fault-tolerant controller [34], and their outputs in terms of compensatory angular accelerations have been shown to possess good detection capabilities [35]. The actuator failures considered are left (L) or right (R) control surface (which can be stabilator, aileron, or rudder) locked at trim position plus 8 deg.…”

Simulation Environment for UAV Fault Tolerant Autonomous Control Laws Development

“…Two types of aerodynamic control surface failure are modeled: jammed surface and physically damaged surface (Perhinschi et al, 2008).The first failure type corresponds to an actuator mechanism failure and results in a locked surface; in fact, at the failure occurrence, the control surface moves to a pre-defined position and remains fixed there. A failure involving a blockage of the control surface at a fixed deflection does not alter the aerodynamic properties of the control surface.…”

Development of a Detection Scheme for Aircraft Engine Failures Based on the Artificial Immune System Paradigm

“…Excellent results were obtained at all levels of the FDIE process for all cases considered [36][37][38] ; however, the simplified modeling of the engine as a thrust look-up table depending on Mach number, altitude, and throttle deflection did not allow for a comprehensive treatment of the propulsion system abnormal conditions modeling and -for that matter -of the propulsion system FDIE. In this paper, the Modular Aero-Propulsion System Simulation (MAPSS) model 39,40 developed by NASA has been linearized and interfaced with a supersonic fighter aircraft model 41 and the WVU motion-based flight simulator to provide the adequate framework for the development and testing of the detection scheme. A variety of aircraft engine actuator and sensor failures were modeled and implemented into this simulation environment.…”