Research On Vibration Online Monitoring And Fault Diagnosis Of CNC Machine Tools

The abnormal vibration of CNC machine tools often accompanies with faults. Inthe machine tools, there are several kinds of vibrations. Determining the fault typeand the source of the abnormal vibrations by online monitoring and diagnosis of thevibration signal is very important for the operation safety of the machine tools andimproving the machining quality and efficiency.In this research, an online vibration monitoring system for the TDNC-SX-TACNC machine tool was built and an intelligent vibration signal conditioning unit wasdesigned. Three studies has been carried out on the proposed monitoring system,including the early fault diagnosis based on the online monitoring of the vibrationsignal, the forecast of the cutting chatter and the recognition of the tool wear.To solve the complex problem in the stochastic resonance parameter adjustmentsin the weak characteristic extraction of the early stage faults, a united parameteradjustment method is proposed. Based on the analysis on the mechanism of thewaveform distortion in the output of the stochastic resonance, a systematic stochasticresonance signal recovery approach is proposed. To achieve the optimal onlinedetection of the vibration and impact signal, an adaptive stochastic resonance methodis proposed, which is optimized by the particle swarm algorithm. The effectiveness ofthe method was successfully demonstrated by its applications in both the vibrationsignal analysis for the cutting experiments and the diagnosis of the mechanicalloosing.Considering the influencing factors such as the characteristics of cutting tools,workpieces and the parameters of the machining process, the stability of cuttingsystem and the chatter prediction problems are studied. The dynamics model of thecutting system with multiply degrees of freedom is built and the stability lobediagram for high-speed cutting is plotted. The influences of the dynamic responseparameters of the cutting system and the machining parameters on the stability areanalyzed. The principle for the cutting parameter adjustments is established to ensurethe maximum removal rate during the cutting processes. The characteristic of the chaos in the chatter vibration signals is analyzed, and an online monitoring andprediction method for the cutting chatter based on the normalized K-S entropy of thevibration signal is proposed. The effectiveness and the accuracy of the algorithm wereverified by the cutting experiments.The online vibration signal monitoring scheme for tool wear detection is built,and the multi-dimensional characteristics of the time and frequency domain areextracted and analyzed. The LPP method is applied for the multi-dimensioncharacteristics fusion and dimension reduction. The characteristics are classified bythe PSO optimized SVM model, and the recognition result of the classifier isvalidated by the5-fold cross-validation method. Experiment results illustrated that,by employing the proposed method, a high recognition rate for the tool wearcondition and a fast response can be achieved.