Hidden Markov model based micro-milling tool wear monitoring

By taking the micro-milling tool wear identification as research object and through considering the possible phenomenon of single edge cutting,Hidden Markov Model(HMM) of tool wear was established.HMM judged whether the single edge cutting phenomenon appeared or not in steady-state cutting condition firstly.Then wavelet packet decomposition was used to extract the cutting force feature.Eight optimal cutting force features were extracted as HMM training input vectors by Fisher linear discriminance.For single edge cutting and two edges alternative cutting of multiple cutting parameters,three different wear stage HMMs were established to identify the actual wear state of tools,and the most suitable recongnition model was determined through Euclidian linear discriminance.The experimental results showed that the micro-milling tools wear state could be accurately identified by HMM,and the accuracy rate was about 85%.