Preliminary study towards the definition of a PHM framework for the SCAS system of primary flight control systems for helicopters.

The Stability Control Augmentation Systems (SCAS) are devices frequently used in rotary-wing aircraft to improve the stability and reliability of flight and hovering under critical operating conditions; such as rescue missions, maneuvering in cramped areas or close to nearly vertical walls, low visibility or in the presence of heavy turbulence. The SCAS system acts via a limited authority actuator on the aircraft's main control linkage, imitating the pilot-induced command into the movement of the helicopter's main flight actuator. The definition of a Prognostic and Health Management (PHM) framework is crucial for these devices, as it would improve safety and reduce the possibility of sudden failures. It would also bring significant logistical and economic advantages, enabling the maintenance procedures to be optimally organized. In this work, a feasibility study is carried out towards the definition of a PHM framework for the SCAS system. It is considered the possibility of employing additional sensors acquirable from the system to increase the number of signals (and thus of fault indicators, i.e. features obtainable) and improve the process of fault identification, classification and prognosis. The following five steps are presented: 1) Overview of the main components and flight systems of rotorcraft and main phases in the design process of a PHM framework. 2) Description of the SCAS case study and its main subsystems (electro-hydraulic servo valve and actuator with centering spring) and theoretical background on the degradation phenomena considered in both the servo valve and the SCAS actuator. Information is also provided on the operational scenario in terms of temperature and pressure variability, and how this affects system dynamics. 3) Analysis of the equations governing the high-fidelity model used for data acquisition of the SCAS system and description of how the seven degradations introduced affect the system parameters. Presentation of the simulation activity and the data augmentation process used to fully capture the solution space of the simulated SCAS system. 4) Presentation of the Feature Selection process: a robust feature set is defined by considering both in-flight mode and preflight checks, as well as additional and traditionally available sensors onboard the SCAS system. A large number of candidate features are studied using different performance indicators: correlation matrix and signal to noise ratio (SNR) allow a preliminary selection; then metrics such as accuracy, expected severity at detection allow a further sorting of the best failure indicators. The non-redundancy between the failure indicators is established through Kullback-Leibler divergence. The classification potential of onboard signals alone is analyzed. Four feature sets at increasing criticality level are tested and evaluated via the confusion matrices of the respective classification processes using Support Vector Machine (SVM). Significant performance improvements are achieved by considering additional sensors, nevertheless, promising results are obtained by combining flight mode signals and pre-flight checks without the use of additional sensors. 5) Presentation of the functioning of the PHM framework: functional examples are provided of the process of fault detection, classification and subsequent prognosis. Here a Particle Filter (PF) technique is used to estimate the hidden state of the degradation.