Model-Based Life Extending Control for Rotorcraft

Abstract

Rotorcraft are subjected to fatigue loads that not only limit the fatigue life of components but also add to their cost. Most of the fatigue-critical rotorcraft components are located in the rotor system, creating challenges for real-time load and structural health monitoring of such components. Furthermore, in forward flight, as the helicopter's main rotor rotates and simultaneously advances, a very complex aerodynamic environment dominated by large dynamic loads is created. Because of the asymmetric air flow past the main rotor, the lift forces each blade generates vary depending on its location. This creates cyclic loading that occurs at the main rotor frequency of rotation (1/rev) and at higher harmonic frequencies (n/rev, n = 2, 3 ,4, etc.) which become important for vibration, fatigue, and forward flight performance. Hence, many components in the rotor system are highly loaded with cyclic loads at multiples of the rotor frequency. In addition, during aggressive maneuvers, the low-duration high magnitude cyclic loads may lead to small amounts of localized damage, for example, localized plasticity, at stress concentration regions. Therefore, it is crucial to develop control strategies that can guard against premature fatigue failure of critical helicopter components to enable component life extension. This research aims at developing real-time algorithms that estimate component level dynamic loads in order to enable real-time load monitoring of critical rotor components and control strategies which alleviate or limit fatigue damage.

A nonlinear helicopter model with 33-inflow states and elastic blade representation is modeled in FLIGHTLAB. The developed nonlinear model gives a suitable representation of the dynamic loads that the rotor system experiences. From the nonlinear model, a first order Linear Time Periodic (LTP) model of coupled body-rotor-inflow dynamics is extracted by performing a linearization about a periodic equilibrium. The LTP model is transformed into a Linear Time Invariant (LTI) model using harmonic decomposition methodology. The obtained LTI model which has 1513 states is used to develop novel schemes for online estimation of rotor component loads.

The fidelity of the 1513-state LTI model is assessed in the frequency domain via comparison with flight test data. A model order reduction approach based on singular perturbation theory is used to reduce the 1513-state LTI model to a 10^{th} order LTI model. The 10^{th} order LTI model retains the physical meaning of relevant states and the fidelity of the dynamic load prediction of the 1513-state LTI model.

Using the reduced order LTI model, two component load limiting strategies to limit fatigue damage are pursued. The first one is based on a receding horizon model predictive control (i.e., Load Limiting Control (LLC) scheme) while the second one is based on active rotor control (i.e., Load Alleviation Control (LAC) via IBC scheme). In both approaches, component life extension is achieved by directly limiting fatigue life usage associated with harmonic loads. In the receding horizon model predictive control formulation, an optimal control problem is formulated where given a desired user-defined maximum harmonic load limit, an estimate of the control margin associated with the component load limit is found and used in the form of pilot cueing/automatic limiting to prevent the component harmonic load from exceeding the maximum limit. In this approach, the use of the reduced order LTI model is twofold. The component harmonic load estimate generated by the reduced order LTI model is used in the detection of limit violation. Furthermore, the reduced order LTI model is used to generate a mapping between the limit and control margins. To assess the effectiveness of this scheme, its integration with a visual cueing system is performed. Subsequently, the resulting architecture is implemented within the Georgia Tech Re-configurable Rotorcraft Flight Simulator to perform real-time piloted flight simulation experiments.

The component load limiting scheme based on active rotor control uses the 10^{th} order LTI model in the synthesis of a higher harmonic individual blade controller (i.e, IBC controller). The IBC controller uses load predictions from the 10^{th} order LTI model to compute optimal higher harmonic individual blade pitch inputs to reduce specific harmonic loads. It is found that the proposed component load limiting scheme via IBC is effective in reducing desired harmonic components of pitch link load at trim but also during maneuvering flight with no impact on the maneuver performance and vibratory hub loads. Furthermore, using the handling quality requirement for small amplitude pitch changes in forward flight, it is shown that the proposed scheme does not cause handling qualities degradation.