Study On Friction Compensation For X-Y Table

Numerically controlled machine tools are the basic equipments to carry out advance manufacture technology. They relate to the country's tactic status of development in the world. Nowadays, high velocity and high accuracy are the developing trends of numerical controlled machine tools. Nonlinear friction inevitably debases the accuracy of servo systems. So, appropriate methods must be carried out to eliminate the frictional effects. Satisfying results only can be achieved at the cost of high control input if we compensate friction with robust methods or traditional PID methods considering it as uncertainties or disturbances. It is urgent to study the friction character and compensate it accurately. There are two types of compensation: model based and non-model based. Non-model based compensation always has limitations. In the model-based compensations, how to identify the nonlinear parameters is an intractable issue. Especially in dynamic model, the variety of nonlinear parameter is neglected, therefore the compensations based on the model lack of adaptability. Firstly, friction character, friction model and the friction compensation methods are introduced. Then, the frictional affections on servo system are studied via simulation based on X-Y table. Some compensation methods are brought forward and simulation results have proved the effectiveness of these methods. 1. An adaptive fuzzy system is employed to learn the nonlinear friction character on line to estimate the friction force and counteract it. Adaptive fuzzy compensation has the advantages of both adaptive control and fuzzy logic and can learn the varying nonlinear relation between the velocity and friction force. 2. Friction is separated into linear part and nonlinear part based on exponential friction model. A much simpler fuzzy system is designed to identify the nonlinear part of friction, which means less fuzzy sets and rules. The robust term in the former method is replaced by an adaptive parameter, which makes the control law simpler. 3. In the compensations based on LuGre model, the uncertainty and changeability of nonlinear steady parameters are taken into account. An RBF neural network is employed to...