Research Of Methods On Fault Analysis Of Advanced CNC Machine Based On Data Mining

Data mining is the prosess of extract the unknown and potential useful information andknowledge form original data which is incomplete, fuzzy, random and contain lots of noisy. Datamining is widely concerns recently years. As an interdiscipline subject, it is related to severaldomains such as data base, Artificial Intelligence, machine learning, statistics, patternrecognition, visualization parallel computing and neural network.Advanced CNC machine is the key production equipment of modern manufacturing, andthe research of dynamic monitoring and intelligent fault analysis is of great significance toimprove manufacturing quality and machine utilization rate. By the cause of the complexity ofthe structural and the variability of the running status of CNC, it is always contain noise whencollect sample dates. We propose a system of fault analysis of advancedCNC based on Roughset.According to the difficulty of collect the CNC machine fault data, In this paper we first givethe method to collect CNC machine fault data based on fault diagnosis knowledge, and discussthe extract method of fault information featuring in time, frequency and time-frequency domain.Then carries the research of Rough set discretization method of fault information based on thecouple of positive domain and information entropy, proposed a new measure of the importanceof the cut point, experimental result shows that the new method better in reduce the number ofcut points and increase the accuracy of classification. Based on the discretization of the decisiontable, there are some attribute reduction methods to be discussed. Among which theCfssubseteval was descriptied in detail. Then with the continue we give the overviews of extractrule methods and discussed the DA method through examples.The series of the methods based on rough set theory proposed in this paper give thefoundation of fault diagnosis of CNC machine.