Research On Monitoring Technique For Tool Wear Based On Cutting Sound Signal

Metal cutting is the most common process manner in mechanical manufacture. It's so essential to monitor the tool condition in the manufacture procedure. With the development of FMS, CIMS and other automation techniques, the new tool wears monitoring system which is more practical and reliable in critical demand.Cutting sound signal is very important in cutting process. It is known that the cutting sound signal can reflect tool wear condition. But how to extract the essential characteristics of cutting sound signals for recognizing is very urgent and important. So it is discussed in this paper.Firstly, we summarize the common analysis methods of sound signal in time field, frequency field and wavelet analysis field. Then we set up the experimental platform and collect the cutting sound signals in different tool wear conditions. Using the wavelet transfer method, the cutting sound signal is separated into different frequency bands. The percentages of each band's energy are extracted to be the characteristics to monitor the tool condition. Meanwhile, a three-layer Back-Propagation artificial neural network is built up to recognize the cutting tool wear. Different hidden layer nodes of neural networks are compared to determine the optimized neural network structure. The simulation results figured out by the trained network prove that the means to recognize the cutting tool wear is effective. Then the tool wear edge image and VB are obtained by using digital image process technique, and the monitoring program is built up. Finally the experiments show that the program to recognize the tool wear condition is reliable.