On The Fault Diagnosis Of CNC Machine Cutting Tool Based On The Immune Nerve System

As the automatic technology of CNC machine is improving quickly, the requirments forthe processing technical performance indicators are becoming more strictly. In order tomake sure the quality of the processing of the workpiece, we need to monitor and analyzethe condition of the cutting tool reasonably. In that case, the troubles of cutting tool thatmay occur in the future can be predicted, the cutting tool can be used more effectively andthen we can reduce the rate of the accidents of the troubles of the cutting tool. As a result,some unnecessary economic loss can be decreased. Artificial Immune System, inspired bybiological immune theory, became a hot research issue in the field of computationalintelligence, following the steps of the artificial neural networks and genetic evolutionaryalgorithm. The artificial immune algorithm combined with the neural network algorithmcan be applied to the fault detection of CNC machine tools, which can predict the fault thatmay occur in the future, and can provide basis for the fault prevention in advance, which isan intelligent forecast in the fault of machinery equipment. This provides an importantbasis for advanced to prevent fault, which is not only meaningful in theory but alsopractically significant in engineering.The dissertation expands on the fault prediction of immune neural system based on thetheory of fault formation of CNC machine, combined with the artificial immune algorithmas well as the neural network algorithm and in the background of the fault prediction ofCNC shoe last machine tool, through the mechanism analysis to fault formation of cuttingtool for CNC machine. Regarding the signal of temperature and vibration produced byrubbing between the cutting tool and the workpiece as the information of the fault situation.By analysis and processing of the characteristics information which collected in some ofnormal operating conditions, Hence, the “self” model space is derived from theaccumulation of the characteristic information of CNC shoe last machine tool. In normalconditions, with the lack of fault samples, the “non-self” space samples is achieved bynegative selection algorithm to select, and training generate the detector sets. Therebyexpanding the scope of the training samples. To generate the detector set and sample set asthe input of the neural network, and then to introduced the immune vaccines into theneural network, thus establish the structure of the artificial neural network. Then through training to completed the fault prediction model which based on immune neural network.Based on the characteristic information of the CNC shoe last machine tool, the damagedegree can be predicted, carrying on the analysis of the untrained sample data and testingthe feasibility of algorithm through simulation experiments. In the simulation experiment,by comparing the abnormal situations under the temperature feature and vibration featurerespectively and the abnormal condition under between the temperature and the vibration,the conclusion can be made that the characteristic information combined temperature andvibration is able to reflect the abnormal condition of the cutting tool appropriately. By theuse of the characteristic information of the known fault condition of the cutting tool, to beanalyzed by simulation testify can further prove the feasibility and accusations of themodel established through the prediction results, whose value of application is practicallyguidance meaningful.