Fault evaluation and adaptive threshold detection of helicopter pilot using bond graph method

  • K Imani
  • A.B. Novinzadeh
  • K.N. Toosi
Keywords: Helicopter Control, Fault Detection and Isolation, Human Fault, Bond Graph

Abstract

Hitherto, in the field of aerospace science and industry, some acceptable results from control behavior of human operator (pilot), are caught using usual methods. However, very fewer research, has been done based on personal characteristics. The performed investigations, show that many of happened faults (especially in helicopter), would result to the loss of natural behavior of system, and eventually some fatal accidents. Therefore, development of tools of assessment of pilot in this dynamical system, is one of the vital necessities. The tools of management of system, should be such that, can show the fault. The object of this paper is assessment of the modelled pilot in a simulator, in presence of standard fly inputs. For this purpose the performance of the pilot for collective and cyclic control of the helicopter, is investigated. The used methods are based on mathematical models and by bond graph modeling method. The existence of uncertainties in the simulator system and the modeled pilot’s behavior, causes creation of fault threshold, for human behavior of pilot. By having these fault thresholds, the improper actions of the pilot, cause creation of behavioral reminders, which can perform the assessment of the pilot. By other speech, using fault detection and isolation (FDI), by bond graph method in state space, the assessment is performed and a non-zero reminder shows existence of fault in the system. Therefore, in this paper there is the novelties for modeling the pilot’s body performance and the helicopter’s systems integrally and designing a fault detection system for it that shows the source of the fault, and obviously it can be useful in aviation industry.

Keywords: Helicopter Control, Fault Detection and Isolation, Human Fault, Bond Graph

Published
2017-07-25

Journal Identifiers


eISSN: 1112-9867