In this research, a laboratory platform which has 2 degrees of freedom (DOF), the TwinudRotor MIMO System (TRMS), is investigated. Although, the TRMS does not fly, it hasuda striking similarity with a helicopter, such as system nonlinearities and cross-coupledudmodes. Therefore, the TRMS can be perceived as an unconventional and complex "airudvehicle" that poses formidable challenges in modelling, control design and analysis andudimplementation. These issues are the subject of this work.udThe linear models for 1 and 2 DOFs are obtained via system identification techniques.udSuch a black-box modelling approach yields input-output models with neither a prioriuddefined model structure nor specific parameter settings reflecting any physicaludattributes. Further, a nonlinear model using Radial Basis Function networks is obtained.udSuch a high fidelity nonlinear model is often required for nonlinear system simulationudstudies and is commonly employed in the aerospace industry. Modelling exercises wereudconducted that included rigid as well as flexible modes of the system. The approachudpresented here is shown to be suitable for modelling complex new generation airudvehicles.udModelling of the TRMS revealed the presence of resonant system modes which areudresponsible for inducing unwanted vibrations. In this research, open-loop, closed-loopudand combined open and closed-loop control strategies are investigated to address thisudproblem. Initially, open-loop control techniques based on "input shaping control" areudemployed. Digital filters are then developed to shape the command signals such that theudresonance modes are not overly excited. The effectiveness of this concept is thenuddemonstrated on the TRMS rig for both 1 and 2 DOF motion, with a significantudreduction in vibration.udThe linear model for the 1 DOF (SISO) TRMS was found to have the non-minimumudphase characteristics and have 4 states with only pitch angle output. This behaviourudimposes certain limitations on the type of control topologies one can ado·pt. The LQGudapproach, which has an elegant structure with an embedded Kalman filter to estimateudthe unmeasured states, is adopted in this study.udThe identified linear model is employed in the design of a feedback LQG compensatorudfor the TRMS with 1 DOF. This is shown to have good tracking capability but requires.udhigh control effort and has inadequate authority over residual vibration of the system.udThese problems are resolved by further augmenting the system with a command pathudprefilter. The combined feedforward and feedback compensator satisfies theudperformance objectives and obeys the constraint on the actuator. Finally, 1 DOFudcontroller is implemented on the laboratory platform.
展开▼
