Study On Wear Condition Monitoring System Of Turning Tools

The technology of tool state monitoring is crucial in advanced manufacturing which can not only improve product quality,or reduce the production cost,but also increase production efficiency.It is an indispensable key technology for realizing automation and intelligent production.However,the technology of tool condition monitoring is not mature enough to be applied in actual processing.In this paper,a tool state monitoring system has been built through the research on the technology of tool condition monitoring,and this system is based on LabVIEW.Firstly,due to the study of wear mechanism of the tool,the wear degree of the tool can be divided according to the wear amount of the tool flank.Analyze the advantages and disadvantages of various monitoring methods.Using the indirect method,vibration and acoustic emission sensors were selected as the signal source,a test plan has been formulated,and a cutting machining test has been conducted.The vibration and acoustic emission signals can be collected under various cutting parameters and tool states through the data acquisition system which is built by LabVIEW software.Secondly,the waveform characteristics in different domains are observed,the relevant characteristic parameters are extracted,and the correlation between the characteristic parameters and the tool wear state is analyzed by the analysis of collected signals in time domain,frequency domain and time-frequency domain.In the time domain,the time-domain statistical feature mean,variance and mean square value of the signal are extracted;power spectrum analysis is performed in the frequency domain;4-layer wavelet packet decomposition is performed on the vibration signal in the time-frequency domain;the acoustic emission signal is multi-resolution in 8 layers.The energy percentage of each frequency band has been extracted respectively to form a 31-dimensional feature vector.Again,Relief-F feature selection algorithm has been used to filter the extracted feature parameters and select the feature quantity which is closely related to the tool state(root mean square of vibration signal and wavelet packet decomposition A4,A6,A11,A15 frequency band;AE signal The multi-resolution decomposition of the D2,D4,and D6 bands),these feature can be formed an 8-dimensional feature vector as an input vector for pattern recognition.after the signal analysis and the feature selection,81 sets of samples had been obtained which can be divided into training samples(54 groups)and test samples(27 groups).The established BP neural network model can be used to train samples,and then test samples can be inputted to view the recognition results.After sample has been trained,the correct recognition rate of the BP network model is 92.59%.Finally,a complete tool condition monitoring system is developed based on LabVIEW software and MATLAB software.The system can complete functions such as data acquisition,data storage and reading,waveform display,signal analysis and tool status recognition.After test verification,the system can accurately identify the state of the tool.