Research On Tool Wear Monitoring Of Milling Cutter Based On Co-integration

Milling is one of the most usual cutting means. Its wear stage influences cutting surface quality and measurement precision of parts directly. Aiming at wear monitoring of milling cutter, parameterized modeling of milling force has been realized by co-integration. Classification of wear stage has been achieved by Intellectual Diagnosis Technologies. The detailed research is as follows:Co-integration of cutting force has been researched deeply. Pointing at the non-parameterized feature extraction without mathematical modeling, co-integration of cutting force has been established by Augmented Dickey-Fuller test, co-integration test as well as estimation of co-integration vector, which may reflect dynamic change of wear stage more clearly.Recognition of wear stage has been realized based on stationary test of co-integration process. Due to periodic change of cutting force, cutting force signals of three wear stages have been substituted to co-integration models of the other two stage, during which new co-integration processes would be obtained. The recognition of different wear stage would be realized by testing stationarity of new co-integration processes. The result shows this method can avoid disturbance of periodic change of cutting force.Different wear stage of milling cutter has been classified based on characteristics of co-integration vector combining with Support Vector Machine. Co-integration vectors have been considered as feature samplings. Artificial Neutral Net BP, Hidden Markov Model and Support Vector Machine have been employed relatively to train the sampling and recognize different wear stage. The result shows Support Vector Machine has better recognizable effect with low dimension and small content samplings.For the research above, monitoring based on co-integration model can reflect dynamic change of wear stage more accurately. Meanwhile different wear stage of milling cutter can be recognized by co-integration combining with Support Vector Machine more correctly. The implementation of these studies has a great significance of miproving recognizable precision, cutting efficiency and surface quality.