Research On Data-driven Wear Monitoring Technology Of Superalloy Cutting Tool

Superalloys are widely used in the manufacture of aero-engine core components for the excellent tensile strength,corrosion resistance and thermal stability at high temperature.This kind of material not only has great service performance,but also has the characteristics of high strength,high hardness,low thermal conductivity,etc..Therefore,it is very difficult to machine.These characteristics lead to problems such as large fluctuation of cutting force and high temperature in the cutting area during the processing of superalloys,which results in severe tool wear during the cutting process,and seriously affects the continuity of processing.With the advancement of material technology,ceramic tools are widely used in the manufacturing process of superalloy parts due to the high hardness and high temperature resistance.The tool wear problem of ceramic tools cutting superalloys has gradually become a key research direction in the industry.In this paper,the experimental research and wear signals monitoring of the tool wear process of ceramic tools are conducted during cutting superalloys.Noise reduction,invalid value removal,signal downsampling and other operations on the tool wear signals are carried out.The processed wear signals is analyzed in time domain,frequency domain,and time-frequency domain.Further,the traditional machine learning method and the deep recurrent neural network model based on the deep learning method are used to extract the features of the processed tool wear signals respectively,so as to realize the monitoring of different tool wear states.Finally,the research content of this paper is verified by the hardware and software system for wear monitoring of superalloy cutting with ceramic tools,and a solution for wear monitoring of superalloy cutting tools is given.Specific research contents include:(1)Build a ceramic tool cutting superalloy wear experiment platform,design a three-factor and three-level orthogonal experiment,carry out machining under nine working conditions,and perform four cuts for each set of parameters to obtain tool wear signals under multi-working conditions.During the machining process,the tool vibration signals and acoustic emission signals are collected,and the wear images of the ceramic tool after each cutting are observed to determine and record the current wear state of the ceramic tool,which is the basis for the subsequent ceramic tool.(2)The improved lifting wavelet noise reduction method is presented to denoise the tool wear signals.The invalid signals is removed when the tool is in the no-load state.At the same time,downsampling is performed on the denoised signals to reduce the amount of subsequent computation and expand the number of experimental samples.Then,the wear signals is analyzed in time domain,frequency domain and frequency domain,and the original high-dimensional characteristic parameters of the wear signals are obtained,and the correlation between different characteristic parameters and the four tool wear states is analyzed according to the calculation results.(3)Based on the manifold learning method,the original high-dimensional features of signals are reduced to obtain low-dimensional manifold features,and the support vector machine method is used to classify different wear states.With the help of deep learning method,a tool wear monitoring model based on deep recurrent neural network(DRNN)is proposed.After comparing the calculation results of the two models,it is found that the deep recurrent neural network model has higher recognition accuracy than the manifold learning method for monitoring the wear states of ceramic tools.(4)Based on the Matlab software,a wear monitoring software system for ceramic tool cutting superalloys is built,which realizes the functions of wear signal display and collection,monitoring model pre-training,online monitoring of wear status,and offline analysis of wear signals.In the actual machining process,the monitoring system is used for signal acquisition and wear monitoring,which verifies the practicability and effectiveness of the system.