Study On Cutting Performance Of GH3128 Nickel-Based Superalloy

The GH3128 nickel-based superalloy is widely used as the key hot-end component in the manufacturing field of aerospace engines and the exposed live parts of navigation equipment.As the GH3128 has the excellent performance such as high strength and hardness in high temperature and high corrosion resistance,it is very difficult to machine.The milling mechanism of Nickel-based alloy GH3128 is discussed by combining experiments with mathematical model.In this paper,the cutting process of the alloy GH3128 will be researched from the aspect of cutting force and quality of machined surface.The cutting force is often regarded as the main reference basis of cutting process design,such as the calculation of cutting power,cutting parameters of reasonable formulation,and the design of machine tools champs.The surface roughness is the most commonly used parameter to measure the quality of machined surface,and also as the guidance of cutting force formulation.In conclusion,the study of cutting force and surface quality is of great significance for the material processing and manufacturing.Due to the lack of cutting parameters for GH3128,the grey correlation analysis and fuzzy comprehensive evaluation method is combined together to estimate the machinability of GH3128.In the process of Fuzzy comprehensive evaluation,theGaussian membership function would be repalaced by constructing a new function which could greatly simplify the assessment process and reduce costs in terms of improved metal machinability evaluation method of the application.Based on the machinability of materials classification,the cutting parameters of GH3128 are reasonably determined by considering the same machinability level of other metals.Carbide Al-Ti coating state slotting cutters were used on milling GH3128 for both singlefactor and orthogonal(Taguchi)design on CAXA tool.Sincethe cutting force signal collected in the experimentwas coupledwith a lot of noise signal,the program that using power spectrum density method and then designing a filter cannot be applied to the filter of nosie signal from the cutting force signal.Hence,the method of combining the power spectrum density analysis and Wavelet transform phase could be used to drop noise signalsfrom the cutting force signal.This way can obtain the low noise of force signal,and finally improve the cutting force credibility.The experimental data of cutting force can be used tocreate the classical linearand machine learning models for cutting force and surface roughness,respectively.The globaltest was used to validate the assumptions of regression model.This behavior could improve the accuracy of the linear regression model.In establishing a machine learning model,we compared the performance of several machine learning models,and then analyzed the importance of cutting parameters with the better models.