Model Prediction-based Solutions To Flight Control Of MCUH

Among many types of multi-rotor unmanned aerial vehicles(UAV), the miniature coaxial unmanned helicopter(MCUH) with the coaxial configuration has gain world popularity owing to some distinct advantages including its compact structure, higher maximum forward speed and higher payload to dimension rate. However, MCUHs are usually unstable, strongly coupled and under-actuated high-order nonlinear systems, which render the flight controller design very challenging. Therefore, studying the relevant problems of autonomous flight controller design is of great scientific significance. This dissertation focused on the theoretical analysis and numerical simulation of flight control of MCUH, and obtained the following results:In the first part of this research, we developed a structured model based on the Newton-Euler rigid body dynamic equations, forces and torques generated from various parts of the MCUH were formulated, aiming to precisely reflect the physical characteristics of the employed platform and thereby enable accurate simulations. This model was later on simplified in consideration of difficulties in the controller design caused by the general high orders of MUCH. The resulting model explicitly accounts for the dynamics of lower rotor and uses a lumped, equivalent model for the upper rotor and stabilizer-bar. In the end, the fidelity of the helicopter was verified by comparing the actuated responses from the simplified model and the full-order nonlinear model in time domain. The responses of the proposed model were consistent with what were generated from the full-order one. Meanwhile, low orders and significant physical meaning distinguished the simplified model from other models illustrated in literatures.The second part of this research aimed to attain stability control of MCUH under near hover and low speed conditions. For the practical purpose, we employed an algebraic approach to synthesize the controllers of attitude and position loops under hover and lower speed conditions. By taking into account of the closed loop control performance, the desired characteristic polynomials of MCUH were configured, and accordingly a simple and robust controller was designed. Considering difficulties in choosing the coefficient of characteristic polynomials, we presented a periodic selection rule to determine the desired coefficients, and the resultant system exhibits better transient responses. And then, by integrating MIMO feedback strategy, the proposed method based on SISO control system was extended to MIMO scene, and the application range was therefore extended. Afterwards, the validity the proposed method was confirmed by performing a comprehensive performance evaluation on the proposed controllers.In the third part of this research, we proposed a NMPC-based control synthesis strategy form the point of view of engineering implementation, in view of the constraints on control and inputs for stability flight control of MCUH. Utilizing the dynamic difference both in guidance loop and attitude loop, the classical single integrated NMPC-based control loop was divided into two sub-loops: the NMPC-based guidance loop and linear attitude control loop. In the guidance loop, the proposed NMPC design method worked in a piecewise fashion to reduce the computation burden and to increase the available time for implementing online optimization. In addition, a robust controller with attitude inner loop augmentation was constructed using the algebraic approach developed in Chapter 3 which compensated the low bandwidth of the outer loop(NMPC control loop). Compared with the classical NMPC-based control strategy, the control strategy proposed here has clearly less computation cost. The stability of this control strategy was then verified by theoretical analysis and numerical simulations.At the last body of this research, two optimization-based control schemes were proposed to solve corresponding problems presented in classical flight control including, trajectory tracking and path following. Again, MPC was applied in both designing schemes due to the constraints.In context of the trajectory tracking control, a fast model predictive control design was developed. Utilizing the differential flatness of MCUH, a linear time varying(LTV) model was employed to approximate the nonlinear dynamic behavior of MCUH, and further on applied in the conversion from classical NMPC schemes to LTV-MPC schemes for output trajectory tracking, resulting in reduction of the complexity of the problem solving. Afterwards, the investigation on the traditional trajectory tracking of asymptotically references was extended to time–varying tracking problem. The feedback control law and Lyapunov function were explored to ensure the exponentially asymptotically stable of the trajectory tracking error system. In addition, state dependent end penalty was employed to ensure the convergence of tracking error. Finally, the quickness and feasibility of the proposed method were verified by comparing the simulation results generated from the developed algebraic approach method and the piecewise MPC method.Besides, we designed a sample-data continuous time NMPC controller to solve aforementioned path-following problem. In contrast to previous approaches which depend on the description of path as level sets, a parameterized reference path was employed here. The core idea was using an augmented system, which consists of path error dynamics and path pa-rameter dynamics, to describe and analyze the path-following problem. Therefore the main advantage of this proposed approach is that the path stabilization issue was directly dealt. To start, by allowing an arbitrarily small path following error, an auxiliary controller was utilized, with the help of Lyapunov technique, the problem of stability control of augmented system was then reformulated as a NMPC problem due to input constraints. From the stability analyze perspective, the control Lyapunov function was used to design both the terminal set and terminal cost to ensure sufficient convergence as well as asymptotic stability of the aforementioned NMPC problem. The optimization solutions for NMPC problem at every sampling instant were body velocity and angle velocity, which could barely be applied as real system inputs of MCUH. Therefore, nested attitude-velocity loop control architecture was redesigned following classical loop-shaping and dynamic inversion techniques, which was then used to transform the predicted velocity into suitable servo inputs in a more deliberate fashion. Finally, the feasibility of the proposed control scheme was validated by the simulation results.In conclusion, all the results denoted in this thesis suggested that our developed model prediction-based solutions to flight control of MCUH are feasible and represent significant progress towards optimizing control performance. The proposed methods can elegantly deal with constraints of the system on one hand, and predict and optimize its further behavior on the other hand. Therefore, it provides excellent guidance for the controller design of helicopters, and even for other nonlinear systems with constraints.