Research On Tool Condition Monitoring Technology Based On Deep Convolutional Recurrent Neural Network

With the continuous development of milling technology towards high speed,high precision and high quality,the tool wear has become a key factor that affects the dimensional accuracy,surface quality of the workpiece,processing efficiency and production safety in the milling process.Therefore,the tool condition monitoring technology Research is crucial.In this paper,the wear state of the tool during milling is taken as the research object.Using the vibration signal in the machining process,a variety of traditional machine learning methods are used to learn the tool wear characteristics to realize the recognition of the tool wear state.The specific research is mainly in the following aspects:Firstly,combined with the different wear states of the tool,using the complex data information contained in the vibration signal,the original high-dimensional features are extracted from time-domain analysis,frequency-domain analysis and wavelet packet technology.Feature reduction by utilizing MLLE algorithm,it can reduce redundancy and irrelevance among multiple feature parameters effectively.The low-dimensional features are finally fed into XGBoost classifier which using GA to optimize parameters to recognize the tool state.The experimental results verify the feasibility and effectiveness of the model in tool state identification..Secondly,in order to improve the inefficiency of feature engineering based on the raw data.The DenseNet network and BILSTM network are used to extract the spatial and temporal features adaptively,and map the features of the vibration signal to the tool wear state.This thesis puts forward a CRNN tool state recognition model based on attention mechanism,and applies the model to recognize the cutting tool state,Experimental results show that the model has higher accuracy in tool state recognition tasks.Finally,aiming at the impact of tool wear on machining precision,the relationship between tool wear and vibration signal is established by CRNN model,the tool wear amount is recognized in real time,and the segmented tool wear compensation model is used to compensate the tool wear in the milling process.The error guarantees the precision requirements of the workpiece being milled.