Tool Wear Monitoring Based On Acoustic Emission Technology

The cutting tool is a direct executive in the cutting process, which directly affectsthe processing quality, processing costs and productivity and so on, therefore, the studies of real-time monitoring of tool wear has far-reaching significance to improve the market competitiveness of products.Acoustic emission technology is a non-destructive testing techniques rising inrecent years. This paper in which the tool condition monitoring system is studiedexplores the process and reasons of tool wears, and use the acoustic emission signals tomonitor the tool wear states. It also designs a basic scheme of tool wear state system,according to this scheme to establish an experimental platform which is based on theacoustic emission monitoring system, and achieves the signal acquisition, analysis andprocessing.In C++Builder environment, the collected acoustic emission signals of cutting tool under different wear states are in time domain analysis, frequency domain analysis and wavelet packet analysis. After using wavelet packet to decompose the tool’s acousticemission signals, it’s shown that characteristic frequencies in the bands which is closely associated with tool wear state are majorly distributed between section3and12, andthe energy of these10bands can be used as the characteristic parameters to judge thewear of the state.In the article the state recognition algorithm of BP network are studied experimentally and by using the extracted characteristic parameters a nonlinear mapping relationship connected with the tool state wear. It also discusses the hidden layers of the network, the number of hidden layer neurons selection and network training, verifies the neural network to the validity of the classification accuracy and speed for tool wear state,and provides a new way to solve the existing monitoring problems of the tool wear state.