Research And Application Of Error Compensation Based On Model-free Iterative Learning

Nowadays most of the error compensation methods are realized by measuring and modeling error sources.There are many disadvantages of those methods,such as the error sources are considered not comprehensively,the model is constructed not accurately and the complex model is poorly transplantable.Futher,many of the existing methods have ignored the problem of mass production,and resulted in a serious waste of data resources.We are now in the era of information,intelligence and data,and effective use of data will be the key to increase the profitability of the manufacturing industry.Aiming at the shortcomings of the existing error compensation methods,we present an error compensation method based on model-free adaptive iterative learning here.This method makes the error compensation by only using the input and output data,taking all the error sources into account.The most important thing is that the method is mode-free,thus this method could solve the difficulties and limitations of the traditional methods.In this dissertation,the model-free adaptive method is firstly applied to the error compensation of the machining path.With intelligent error compensation modules designed,we write the corresponding procedure and build the simulation structure diagram in the simulink.The previous data information is applied to the current processing by an adaptive iterative learning method,so as to improve the processing accuracy.The proposed method in this dissertation is verified by Matlab simulation.Based on the verification of the error compensation method of PID iterative learning,the simulation of data-driven model-free adaptive error compensation is carried out.The learning law of PID iterative learning error compensation method is fixed,so the compensation for interference is poor.But the model-free adaptive error compensation method can adjust the learning law in real time,therefore this method can improve the above drawbacks.In order to further verify the effectiveness of the algorithm,a fuzzy iterative learning error compensation method is designed.The fuzzy algorithm can also adjust the system in real time,however,this method has a lot of uncertainties in parameter setting and selection of fuzzy rules,so we can not guarantee its convergence.To verify the compensation performance in the complex environment of this method,we add step and noise interference respectively to the system.Finally,the above algorithms are tested on the XY platform,the simulation and experimental results have proved the feasibility and validity of the proposed algorithm.The proposed algorithm does not rely on the system model and the calculation is relatively simple,so the method is easy to implement.It's a really good combination and application of model-free adaptive method and machine error compensation.