The machining accuracy of CNC machine tool is affected by machine tool parts,external environment and other factors,so it is necessary to add appropriate compensation parameters to ensure the stability of machining accuracy.In addition,the compensation parameters of different lathes at different times change in real time.Therefore,an intelligent error compensation model was proposed based on association rules and neural network method.Based on the dataset produced in actual production,the dataset was screened by Apriori algorithm;each eigenvalue and compensation parameter were normalized to improve the convergence speed of the data;the neural network model was used to search for the best compensation parameter model for lathes in different situations,so as to construct the best intelligent error compensation model;after intelligent error compensation,the produced objects were recognized with image to analyze whether they met the accuracy requirements.The simulation test results show that the lathe stability is improved by 0.695 and 0.713 for the training dataset and the test dataset respectively.The measured results show that 30 products are carved with the above method,and the accuracy meets the requirements.Therefore,the intelligent error compensation model can improve the stability of lathe processing and the product qualification rate.