AE-LSTM-CNN Model based Tool Wear Monitoring

Effective tool wear monitoring is the key to evaluate the degree of tool wear. It is beneficial for monitoring the normal operation of the five axis CNC machining machine tool, improving the processing quality of the products. However, the most of existing data-driven monitoring methods are unable to meet the high requirements of online monitoring, due to their excessive reliance on manual feature extraction and limited model computing ability. AE-LSTM-CNN based tool wear monitoring method is proposed in this paper, which can break the limit of internal mutual block interference with simple sequence combination feature extraction models. Fist, wavelet transform is adopted to reconstruct the tool signal, reducing the dimension and denoising the time series. Then, the reconstructed signal data is respectively fed into the AE-LSTM network to extract feature vectors of temporal signals. CNN networks extract feature vectors of different signal spatial distributions. Finally, the accurate prediction of tool wear is realized through full connection layer output. The verification results indicate the rationality and accuracy of this method.