| The natural unstable two-wheel vehicle is a special kind of wheeled mobile robot. Its dynamics equations are multi-variable, serious unstable, coupling, time-variable, uncertain parameters, non-linear and multi-order. The dynamic of the system is constrained by the nonholonomic kinematics condition, therefore a multitask controller is needed. Such system put forward great challenge for controller. It's also a useful set for testing various control theories. Two-wheel vehicle is a set on which we can study controlling of the systems with uncertain parameters,non-linear systems,adaptive systems and intellectual control. As a convenient vehicle, two-wheel self-balanced vehicle was also valuable.As a complicated non-linear system, the natural unstable two-wheel vehicle is hard to gain good performances with the use of traditional control theories. The first part of the paper, fuzzy control,neural control and fuzzy-neural control were studied. And we find that an intellectual controller would be suitable for this system. At the second part, the system structure and the non-linear mathematic model are given out and linearized at zeros. Then we designe a state-feedback controller in the linearization area based on the analyses of its stability and controllability. At the third part, state-feedback controllers are designed at zeros and two symmetrical spots which near the zeros for the non-linear vehicle model. And then the data from the controlled system is used in the adaptive neura-fuzzy inference system to get the fuzzy controller. But the fuzzy logic which is gained by adaptive neural-fuzzy inference system is static in a short time. Therefore a dynamic fuzzy controller is introduced and uses to balance vehicle. At last, we simulate the three controllers and compare their performances. The result shows that the performance of dynamic fuzzy control system is the best, the adaptive neural-fuzzy take second place, the state-feedback controller is the worst one. |
PDF Full Text Request