Research On Simulation Analysis And Condition Recognation Technique Of Warp Deformation In FDM 3D Printing

The subject of the thesis is based on the natinoal fund project "Research on Theory and Method of Acoustic Emission Monitoring of Fused Deposition Molding 3D Printing" and further carried out on the basis of completed project achievements.In this paper,the most common defect mode-warpage deformation in the FDM 3D printing process is the object of study.The generation mechanism and influencing factors of warpage deformation is discussed with the aid of Ansys.Combining with the AE techology,the method to indentify the wrap state of the part during the process of FDM 3D printing based on feature extraction,cluster analysis and pattern recognition has been researched and constructed.The main content of this paper are:(1)The working principle and process of FDM 3D printing were analyzed,and the parameters and their functions were deduced.The mechanism and influencing factors of warp deformation during FDM 3D printing process.Elaborate the measuring method of the part deformation state in FMD 3D printing process,and led the method that use AE signal to represent the warp state of parts in printing process.(2)The heat transfer process of FDM 3D printing techology were analyzed,and the simulation model is constructed.The thermal effects of the products during the molding process were simulated and analyzed from the aspects of process parameters,workpiece parameters and typical faults,and the influencing level and trend of warp deformation due to parameters and faults were fitted.In addition,according the idea of orthogonal test,the part state in defferent combination of parameters were simulated,which can provide actual machining process and test for theoretical guidence.(3)A sample preprocessing method based on the local extremum of AE signal were put forward for this subject,and the data is denoised and preprocessed.The concept and principle of self-organized map(SOM)algorithm are studied and used in cluster anylay of warp state.The concept and principle of random forests(RF)were studied,respectively in combination with the proposed sample pretreatment method and principal component analysis(PCA)to build the part deformation state monitoring method in FDM 3D printing process.The principle and algorithm of support vector machine(SVM)were expounded,and the advantages and disadvantages of the two methods of pattern recognition in use of RF and SVM are compared theoretically.(4)Based on the experimental data,the state recognition method proposed in this paper is demonstrated.The experimental results showed that the sample pretreatment method proposed in this paper,combined with SOM algorithm and RF algorithm,With high accuracy in monitoring and recognition,it is suitable for on-line monitoring and recognition of the warpage condition of molded parts in FDM 3D printing.