Research On The Cutting Tool State Monitoring

The tool wear has a significant impact on the machining precision and surface quality of workpiece. It also directly affected the product quality, efficiency and the function of cutting system. Therefore, the research on the cutting tool wear state is very important.The cutting forces are obtained by cutting experiments, the feature extractions which reflect tool wear are abstracted by wavelet analysis from cutting forces. Based on neural network, the research on the tool wear state is carried out.The cutting forces are the most important feature in the cutting, there have nearly relation between tool wear and cutting forces, so cutting forces are chose to monitor the tool wear state.The analyses of time domain, frequency domain and wavelet for cutting force are curried out. The relationships between the trend of feature extraction and tool wear are also discussed. The result indicates that it is diffcult for the analyses means of time domain, frequency domain to obtain the feature extraction. The increase of the tool wear, the increasing trend of energy and root-mean-square deviation of wavelet analysis among the feature extraction is obvious. Thus the energy and root-mean-square deviation which are sensitive to tool wear are selected as the best feature extraction.The BP neural network with singledomant can carry out the relation between the feature extraction of cutting forces and the tool wear. The feature extraction which came from wavelet analysis input the neural network with the frame 8-13-l,the neural network is training by training sample, it strikes up the tool wear state model.The result indicates that the system of tool wear state monitoring could be eximious capable of identifing the tool wear state, the system is based on wavelet analysis and neural network.