| Twin Rotor Multi-input Multi-output System (abbr. TRMS) is a laboratory set-up bought from England for the Center Laboratory of Control Science and Engineering Department. It is designed both for theory research and for teaching demonstration. From the control point of view, TRMS is a nonlinear, multivariable system with significant cross-couplings. It can efficiently reflect many important characteristics in the course of control, such as non-linearity, robustness, stability and decoupling, etc. The control strategies we used in TRMS can also be practically useful for designing control systems for general industry processes. The main tasks of the thesis include several aspects as follows. Firstly, some research work of TRMS is summarized in this thesis. Then some methods of nonlinear system control are introduced since TRMS itself is a nonlinear system. Secondly, the mathematic model and the SIMULINK models of TRMS are presented, which include two 1-dof models as well as a 2-dof model. Thirdly, after analyzing the PID controllers of the TRMS models, the strategies of tuning the TRMS PID parameters with Genetic Algorithms and Particle Swarm Optimization Algorithms are suggested in the thesis, since the experience-based tuning method and Ziegler-Nichols optimization method can't deal well with the complex nonlinear system, such as TRMS. Fourthly, the method of system identification with neural network is investigated before using Bp network or neural network based on genetic algorithms to identify the 1-dof vertical model of TRMS. Lastly, the experiments of PID parameters tuning and neural network identification are finished in the thesis by using SIMULINK software in MATLAB. The simulated results suggest that the PID controller will have good track performance after using the parameters tuning method proposed in the thesis, and that the identified mode of 1-dof TRMS can also effectively reflect the input-output characteristics of the system.
|
PDF Full Text Request