Research on Adaptive Neuron-Fuzzy Control for Self-balancing Two-wheeled Vehicle
|School||Xi'an University of Electronic Science and Technology|
|Course||Control Theory and Control Engineering|
|Keywords||Two-wheeled Self-balance Vehicle Fuzzy Linearization Fuzzy Inference System (FIS) Optimal Self-adapting Fuzzy Controller|
In recent years, with the further research of mobile robot, its application fieldbecomes more and more extensive, but the environment it facing becomes more andmore complicated. Gennerally speaking, any mechanical system with wheeled mobiledriving device belongs to the category of mobile robot, which has a broad applicationbackground, so it attracts much attention in the domestic and foreign control field.Two-wheeled self-balance vehicle is wheeled mobile robot which is unstablenaturally. After dynamic analysis, the system is multivariable, time-variant, closecoupling, nonlinearity, parametric uncertainty, etc. Its movement environment iscomplicated and motion equations are non-integrity constraint, It need implement pathplanning and path tracking, while finish self-banlance. So its control is a complex task.The reseach of two-wheeled self-banlance vehicle can effectively reflects many typicalproblems existing in control field, such as non-linearity, robustness, stabilization,follow-up and tracking. Therefore, two-wheeled self-balance vehicle is the idealplatform to study all kinds of control theory problems, and also a typical equipment oftesting the abilityof different control methods and strategies.This paper focuses on the further optimization of mechanical structure fortwo-wheeled self-balance vehicle. Improve the performance by allocating hardware andsoftware resources rationally. On the basis of analyzing the vehicle’s structure andelectronic equipments’performance parameter, we obtain self-balance vehicle’sboundary condition of the controllable angle, which provides significant foundation forvehicle’s performance improvement. By dynamic analysis and kinematical analysis, wefirstly use tradional PID to control the vehicle, but it failes. Then we design twofeedback linearization controllers. One is based on pole placement, the other is based onquadratic optimal control. And we compare the control performance of two controllers.On the condition that controlable is limited, both controllers are not perfect to controltwo-wheeled self-balance vehicle. According to fuzzy linearized theory, we design poleplacement self-adaptive fuzzy controller based on adaptive network-based fuzzyinference system.The resule show this controller can control the vehicle in a great angle,but system state variables oscillate near the banlance point, which result in lowprecision control. Lastly, we design an optimal self-adaptive fuzzy controller, which canachieve control of self-balance vehicle in agreat angle and the control precision are alsoimproved.