Method And Application For Monitoring Abnormal Vibration Of Spindle System In Machining Center

As a key component of the machine center,the spindle has a decisive influence on the machining accuracy and productivity of the machine center.With the increasing requirements for spindle reliability,life and maintenance costs,condition monitoring is very important.Periodic or continuous monitoring of the operating state of the spindle can prevent abnormal conditions such as components wear,overheating and accidental damage,timely identify the spindle condition,detect early signature of failure,eliminate potential failure,and realize the intelligent maintenance of the equipment.This paper mainly studies the monitoring method of abnormal vibration of the spindle system of machining center.The main contents and results are as follows:(1)Signal acquisition of abnormal vibration and analysis for spindle system.The principle of spindle state signal selection is introduced.The selection and optimal placement of vibration sensor are analyzed.The acquisition method of numerical control system signal is studied.The signal analysis method is studied for feature extraction of the spindle condition signal.(2)Fault diagnosis of spindle bearing based on vibration signal.Aiming at the non-stationary vibration signal generated by the spindle bearing wear,this paper proposes a feature extraction algorithm based on S-transform combined with singular value decomposition,and a feature extraction algorithm based on wavelet packet energy.The data set was classified and identified by neural network,K-nearest neighbor and multi-class SVM,respectively.The optimization results of S transform combined with multiclass support vector machine are obtained.(3)Abnormal vibration based detection method for chip in spindle and chip tied in tool.A feature extraction algorithm is proposed based on fundamental frequency energy for detecting the chip inspindle chip and chip tied in tool at the same time.An ID3 decision tree algorithm based on fundamental frequency energy and wavelet packet feature offset is presented for distinguishing different abnormal modes including chip on end face,chip on cone,and chip tied in tool.The decision tree model was trained based on both abnormal and normal modes.The test results show that the decision tree model has fast recognition speed and high recognition accuracy,which can be used as a preferred detection classification method.(4)Spindle health assessment method.A method for evaluating the spindle health based on the wavelet packet energy offset is proposed.The method is tested under the condition of the spindle bearing wear,the normal working condition of the spindle,chip in the spindle and the tool tied in tool.The results show that the method can effectively diagnose the spindle health condition.The chip in spindle has no influence on the assessment of spindle health.The assessment method can be used in the intelligent maintenance of the spindle system in the machining center.