The Design Of Iterative Learning Controller For CNC Machine Tools

Making machine intelligent is the common ideal in industrial control field.In the face of repeated processing tasks,if CNC(Computer Numerical Control)machine tools have the "self-learning" function and can use previous error information to guide subsequent processing,the machining error will decrease.However,the existing CNC machine tools can only treat batch production as a standard machinery one by one.Therefore,prior processing information can't be used,and after a series of complex process,the finished product rate is low.To solve above problems,this paper proposes the idea of self-learning controller for the existing NC machine tools.In many self-learning algorithms,based on the principle of easy and convenient application,we select the iterative learning algorithm.And then,the total controller structure is PID type iterative learning controller with disturbance observer.The process is divided into four steps.They are the modeling and identification of numerical control machine tools,the structure design of self-learning controller,the parameter optimization and the effect verification.In depth analysis of the iterative algorithm,we found that most researches stay at the simulation and half-hardware simulation stage.So we sum up two major difficulties for iterative algorithm's application.First,theory of this algorithm is mostly based on the open loop iteration,which is difficult to guarantee the stability of our system.Second,even the convergence condition is meet,error decreases first and then increases during the iteration process.For the first difficulty,the closed loop iterative structure is selected by considering the convergence precision,the speed and the achievable complexity of the iterative algorithm.At the same time,because the algorithm can only suppress repetitive disturbance,we add a disturbance observer to inhibit non-repetitive disturbance.For the second difficulty,we proposed conditional start and stop iterative mechanism,which is very convenient applying to practical CNC system.In the three axis engraving and milling machine tool verification,for butterfly trajectory tracking,the error under controller we designed in this paper can be reduced to 5.705 ?m after ten iterations.Contrast to optimal parameter self-tuning based on simplex method,error is 33.4 ?m.In the existing iterative learning books and literature,most of them use lots of formula to explain the principle.For the beginner,it's too difficult to understand.In this paper,we combine these theoretical knowledge with numerical control machine tools.Based on the practical application,we describe the detailed process of controller design with the most simple language.This method is written to CNC machine systems,realizing the application value of the algorithm.