| The straight line driving motion as a new driving style is widely used in the actual production, such as the spindle movement of the numerical control machine, the movement of the degree of freedom(DOF) of the robotic manipulators, etc. Therefore it has become the hot issues of the research of control engineering recently which contribute to its target control research fast, agility and high accuracy.According to the characteristics of the straight line motion system, the following work is carried out after the full consideration of the driving link flexibility, the complex uncertain load and the friction characteristics. Comparing least square method, impulse response method and the neural network approximation method when taking the Stribeck friction model as the research object, using genetic algorithms to approach the parameters of the friction, in which can track the real friction curve perfectly.During computer simulation, regarding the linear motor-spring-mass block – spring-mass block system as equivalent model. Firstly, the tracking controller is constructed to follow the reference signal based on sliding mode control theory. Apart from that, BP neural network in sliding mode control is used for uncertain term approximation, what is more for optimizing the control input that related to further improve the control precision, to reduce the peak input control. Then in this paper, design a state observer as the velocity cannot be directly measured during feedback routine, accordingly good results are obtained.Both controller and observer which mentioned above are verified by the virtual simulation software ADAMS. Meanwhile the experimental task is also carried out on the experimental platform based on DSP controller and the fundamental data of the parameter identification in the friction model is obtained. |
PDF Full Text Request