| In the metal cutting processing, CNC machine tool failure is mainly caused by thefailure of the cutting tool. If not timely and effective monitoring fault, there will be thephenomenon of the production process suddenly stopped, causes the waste ofmachining and the damage of machine tool, in addition, unreasonable estimate theservice life of the cutter, will depress tool usage and increase the pressure of toolchanger in processing. The research of the real-time on-line monitoring technology ofCNC machine tools are beneficial to improve the trouble-free operation efficiency ofCNC machine tools, reduce the probability of the workpiece scrapped and equipmentdamage.This paper bases on milling cutter of CNC machine tools as the research object,on the basis of the comprehensive analysis of domestic and foreign research presentsituation and the wear mechanism of the tool wear state s. For single sensor has thedefects such as single information type, imited information capacity and stronganti-interference ability, the method of multiple sensor monitoring the tool wear stateof multi-source information fusion by vibration sensor, pressure sensor, vision sensorsand dynamometer are presented, and the multiple diagnosis model is established basedon artificial neural network to realize multiple model of decision level fusion.At thesame time, with the aid of the LabVIEW and MATLAB software programming, a setof perfect cutter wear condition monitoring system is established, in order to identifydiagnostic studies of the cutter wear condition and realize real-time online monitoring.By research the mechanism of the CNC machine tool wear and the wear types,determine the initial wear, normal wear and sharp wear as the tool wear conditionmonitoring and diagnosis; Through the analysis of commonly used monitoring method,design a milling cutter testing scheme by the methods such as the vibration signals,acoustic pressure signal, cutting force and workpiece surface texture, and based onLabVIEW and MATLAB software programming, design the signal collection andprocessing system for wear conditions of a milling cutter. In order to reduce the test frequency and uncertain factors of test process，designsnine sets of scheme for the milling cutter cutting three elements and simulationresearch based on the finite element software of DEFORM-3D. According to thelength of the service life of cutting tools and extreme difference analysis method,determine the optimal test plan which is the cutting speed of100m/min, each toothfeed of0.1mm, cutting depth of0.8mm as a final dry milling test the parameters ofthe scheme. At the same time, it is the foundation for the subsequent patternrecognition and decision fusion system of building.The time domain, frequency domain and the ensemble empirical modedecomposition(EEMD) of time-frequency domain analysis are studied, characteristicvalue of acquisition to the original signal are extracted.Then by the method of kernelprincipal component analysis (KPCA) to deal with the original characteristics; Finallyconfirmed the35sets of most can reflect the characteristics of the wear conditions of amilling cutter as an effective combination of the characteristics, and then input theminto the three kinds of intelligent network model for pattern recognition research.Finally, through establishing the genetic BP, RBF, SVM network modelcompleted preliminary integration of the wear conditions of a milling cutter; By fuzzyset theory to improve the evidence theory fusion principle a nd model, put forward anew DST improvement method and applied to the milling cutter wear conditionmonitoring of multiple model decision level fusion system. The tool wear conditionmonitoring and diagnosis system which is based on multi-sensor information fusion isestablished, and has obvious advantages in the diagnosis of speed, accuracy andstability, can effectively monitor tool wear state. |
PDF Full Text Request