Feature Extraction Of Signal And Tool Condition Monitoring Of High Speed Milling

The development of manufacturing technology is an important index to measure a country’s industry developed degree, as well as the country’s economic development level. High Speed Milling technology occupies an important position in the advanced manufacturing technology because of its high speed and high efficiency and it can obtain higher machining precision and better surface quality. At the same time, compared with the traditional milling technology, the machining of cutting tool of high speed milling has shorter lifetime. The problem of tool wear has became a great limit to the further development for high speed milling.Tool condition monitoring can significantly improve the usage rate of cutting tool, avoid the result of the cutting tool problem of lower efficiency, mechanical parts unqualified and even machine accidental damage, so there are great theory significance and practice value to take research for taking effective monitoring tool wear condition. Based on the high speed milling cutter wear condition as the research object, mainly from the monitoring system based on sparse decomposition theory of signal acquisition, feature extraction and state judging aspects were studied as follows:(1) In order to get cutting tool state information data, the system signal acquisition scheme was explored. Firstly, three aspects as reasons of tool wear, the types of tool wear and tool wear process had been taken research of. Based on the research, commonly used signal acquisition methods were researched, and vibration signals of cutting tools are chosen as the analysis object.(2) In order to get the essential characteristics of vibration signals, vibration signal features were extracted. Feature extraction methods as time domain, frequency domain and time-frequency domain are used and the feature vectors are obtained. All the feature vectors were screened based on correlation analysis and 10 groups of them as the highest correlation coefficient vectors were selected as the basic data of the experiment.(3) In order to accurately distinguish tool condition and implement effective monitoring of the cutting tool, sparse decomposition theory was used to deal with the signal. The sparse decomposition of related mathematical theories were studied, the processing algorithms based on sparse decomposition were designed, the sparse decomposition over complete dictionary was build, tool condition for the effective judgment was made combine with Euclidean distance. Contrast experiment based on the neural network was taken. Results of the experiment show that sparse decomposition theory can be used in the criterion for tool wear state and the system can realize effective monitoring of cutting tool.