Robust actuator controller for active-truss-based morphing wing.

Abstract

The active-truss-based morphing wing (ATBMW) is a new type of smart structure, which is more efficient than airfoils with conventional control surfaces. However, the sophisticated ATBMW framework and large numbers of actuators make it difficult to obtain the overall structural dynamics for controller design and inconvenient to tune actuators on board. Our research therefore aims to develop an actuator-level control scheme to simplify the process of controller implementation on ATBMWs so that the above problems regarding controller design and on-board tuning can be bypassed. The proposed control scheme is based on the concept of unknown-input estimation and compensation in a servomechanism. A new unknown-input estimator (UIE) is developed and integrated with a Linear-Quadratic-Gaussian (LQG) controller to provide enhanced compensation of uncertainties. By doing so, the resultant controller can be designed and tuned simply using the dynamics of the actuator, without the necessity to know the dynamics of the entire wing structure. Existing techniques for estimating unknown inputs to a system require at least one or more of the following: detailed knowledge on unknown inputs, derivatives of measured outputs, inversion of plant dynamics, constrained state observer design, parameter optimisation (global optimum not guaranteed), or complicated designs. The new UIE developed in this thesis is exempted from the aforementioned limitations and features a simple structure and straightforward design. To validate the proposed UIE-integrated LQG controller, an ATBMW prototype with 5 linear actuators is built. For comparison, a PID controller is introduced in both simulations and experiments. Both types of controllers are designed using two sets of models obtained via system identification: one set represents actuator dynamics only, while the other set includes wing structural dynamics. In simulation study, system sensitivity and stability robustness are firstly investigated against parameters associated with the UIE component, with guidelines for designing the proposed UIE-integrated LQG controller validated. The mechanism of unknown input compensation is then demonstrated by dividing unknown inputs into exogenous disturbances and internal uncertainties and examining the two situations separately. Compared with a standard LQG controller, the UIE-integrated LQG controller shows enhanced capability in rejecting unknown inputs. Lastly, the UIE-integrated LQG controller is implemented on all the 5 actuators in the presence of only internal uncertainties, and compared with the PID controller. Superior performance of the UIE-integrated LQG controller over the PID algorithm is observed in simulations. In experimental study, wind tunnel tests were conducted to further validate the efficacy of the UIE-integrated LQG controller under both aerodynamic loads and modelling errors. The performance of the UIE-integrated LQG controller designed according to actuator dynamics is closely comparable to that of its congener based on wing structural dynamics, and both outperform the PID controller. In conclusion, the new UIE is capable of effective estimation of unknown inputs. The UIE-integrated LQG controller has an enhanced capacity to compensate a wide class of unknown inputs including exogenous disturbances and internal uncertainties, and meanwhile the ease of design is maintained. The most significant merit of applying the proposed controller on an ATBMW is that the implementation of actuator controllers is considerably simplified despite the complexity of the ATBMW framework. The controller can be based on actuator dynamics only, and can be tuned on individual actuators before the actuators are assembled on the wing. Therefore, the process of controller implementation is free from structural coupling constraints, and there is no need to obtain wing structural dynamics for controller design and to further tune actuators on board. Beyond the merits mentioned above, the proposed controller has broader significance in the following two aspects. Firstly, it provides a unified solution to simplifying actuator controller implementation on ATBMWs despite the variations and complexity of ATBMW structures, and is thus significant to successful realisations of a wide range of promising ATBMW concepts; Secondly, the enhanced capacity of disturbance rejection is crucial to aerodynamic improvements achieved by ATBMWs as it ensures reliable performance of wing morphing in the presence of unmeasured and unpredictable exogenous loads.