Research On Prediction Of Tool Wear In CFRP Drilling Based On Improved Whale Optimization Algorithms

Carbon Fiber Reinforced Plastic(CFRP)has excellent properties such as high specific strength,large specific modulus,good rigidity,high temperature resistance,corrosion resistance,wear resistance and so on.It is widely used in aerospace and automotive manufacturing fields.Due to the characteristics of anisotropy and low interlayer bonding strength,the tool wear is severe during drilling,which seriously affects the quality of the surface and aggravates the hole defects peculiar to composite materials such as burr,tear and lamination.In this paper,the machining process of CFRP drilling is taken as the research object,and the prediction of tool wear of polycrystalline diamond(PCD)twist drill,carbide brad-spur drill and carbide dagger drill are carried out based on the signal characteristics of piezoelectric dynamometer,triaxial and temperature sensor,a prediction model of tool wear combining artificial neural network and intelligence optimization algorithm was established.The main work is as follows:(1)The hole machining defect and tool wear mechanism in CFRP drilling and the advantages of multi-feature fusion are described.CFRP drilling wear experiments under multiple types of tools were carried out,monitoring data and tool wear data of multiple sensors were obtained,the wear curves of various tools were drawn,and the wear values of various tools in different wear states were set according to the actual situation,which provided a basis for data analysis and prediction model construction in the following paper.(2)Through single factor wear experiment with variable cutting parameters,the influence of tool structure on wear characteristics and hole quality was studied.It was found that the drilling temperature and the wear value of the back surface of the three kinds of tools have the same trend with the change of cutting parameters.With the increase of the number of drilling holes,the surface roughness of the hole wall and the delamination factor at the exit of the hole caused by the three kinds of tools show an increasing trend.In the machining effect of the surface roughness of hole wall,the brad-spur drill > dagger drill >PCD twist drill;For the inhibitory effect of the delamination factor at the outlet of the hole,the dagger drill > PCD twist drill > brad-spur drill;For fiber burr shear effect,brad-spur drill > PCD twist drill > dagger drill.(3)The method of sensor feature extraction and analysis was studied.65 features were extracted from multiple sensor signals,and combined with cutting speed,feed speed,hole wall surface roughness and delamination factor of exit,a77-dimensional feature space was constructed.KPCA was used to reduce the feature dimension and redundancy,and to weaken the influence of noise.Pearson correlation coefficient was used to prove the validity of fusion features.(4)Aiming at the shortcomings of the Whale Optimization Algorithm(WOA)in terms of convergence accuracy and speed,an amendatory Whale Optimization Algorithm(AWOA)was proposed by combining opposite-learning and adaptive inertia weight.The standard test function was used to evaluate its performance,and the results show that the AWOA algorithm has faster convergence speed and better convergence precision.(5)The tool wear state prediction model optimized by AWOA was used to predict and classify the tool wear state,and the validity of AWOA-BP model was verified.The experimental results showed that the prediction accuracy of AWOABP model is higher than that of some classical algorithms.The tool wear state can be accurately predicted during CFRP drilling.