On-line Detection Of Additive Manufacturing Defects Using Laser-induced Breakdown Spectroscopy

In recent years,with the rapid advancement of the new round of scientific and technological revolution and industrial reform,advanced manufacturing technology represented by additive manufacturing has made rapid development.The research and application of additive manufacturing which plays a vital role in China’s national economic and social development is spreading in high-end equipment manufacturing and high-precision science and technology development fields,such as aerospace,national defense industry,electric power industry,automobile industry,medical equipment and mold manufacturing.The additive manufacturing technology adopts the digital discretization and stacking principle.Metal parts are manufactured through point by point processing,line by line connection and layer by layer stacking of materials.Therefore,this manufacturing technology has the advantages of high machining efficiency,short machining cycle,saving raw materials,simple and high precision forming process,direct forming into arbitrary complex shape structure and so on.However,due to the complex physical mechanism of the interaction between laser and materials in the machining process,different kinds of defects such as cracks,holes,inclusions,poor fusion and balling may appear inside or outside the metal parts,which seriously affects the service performance and the life of the parts.The implementation of on-line inspection of additive manufacturing defects is very important to improve the quality of additive parts.In this paper,SLM metal component defects were on-line detected and identified using Laser-induced Breakdown Spectroscopy(LIBS),which is an in-situ and real-time nondestructive testing method.This paper focuses on the construction of LIBS on-line detection device for defects in SLM additive manufacturing process.The plasma spectrum of molten pool is collected and analyzed through experiments,and the real-time detection and high-precision identification of defects are completely combined with machine learning algorithm.The main research contents are as follows:(1)It has built a LIBS on-line detection device for defects of SLM metal additive parts.On the premise of ensuring that the SLM additive manufacturing process is not affected,combined with the characteristics of processing technology,coaxial double optical path LIBS optical fiber detection probe,three-dimensional lead screw feed position servo drive and other devices are designed.The coaxial double optical path LIBS optical fiber detection probe introduces the pulsed laser beam into the SLM chamber by using the optical fiber transmission method,and transmits the plasma information generated by the excited molten pool to the detector coaxially.Optical elements such as convex lens and reflector mirror are installed in the probe cavity to provide an optical path for laser and plasma transmission.A three-dimensional ball screw feed servo drive mechanism was installed directly above the SLM powder cylinder,which is used to drive the coaxial dual optical path LIBS optical fiber detection probe to complete the stable operation in the SLM working chamber and the accurate positioning of the molten pool focus.At the same time,it has completed the selection,analysis and parameter optimization of experimental equipment.The on-line detection target of defects in SLM additive manufacturing process can be realized through the on-line detection processes such as precise positioning of LIBS laser pulse of the focus on molten pool,introduction and induced excitation of plasma by laser light source,collection and transmission of plasma spectrum information,timing control of on-line detection system,plasma spectrum analysis and defect identification.(2)Based on the dynamic modeling of three-dimensional ball screw feed position servo driving mechanism,two platform dynamic control schemes based on PID controller and linear auto disturbance rejection controller are proposed.Through Matlab/Simulink simulation environment,under the conditions of no disturbance,triangular wave or sine wave disturbance,the tracking effects of two control strategies for given trajectories of straight line,circle and "8" shape are compared and analyzed,and the trajectory tracking performance of the system is verified by experiments.The research show that the linear ADRC does not depend on the accurate mathematical model of the three-dimensional ball screw feed platform,and it is easy to eliminate the nonlinear and uncertain effects of mechanical friction,so as to realize inter shaft decoupling and system disturbance compensation.(3)The AC permanent magnet synchronous servo motor is used to drive the three-dimensional ball screw feed mechanism to complete the movement speed control,trajectory control and position tracking control of LIBS optical fiber detection probe in the working chamber space of SLM additive manufacturing.The research focuses on correcting the current loop,speed loop and position loop of the position tracking servo system respectively.The simulation and experimental results show that the LIBS optical fiber detection probe driven by the three-dimensional ball screw feed position AC servo system has good rapidity and high positioning accuracy,and can realize fast and accurate tracking of SLM molten pool.(4)Three machine learning algorithms,k-nearest neighbor algorithm(KNN),support vector machine(SVM)and random forest(RF),were used to identify the defects of SLM added parts.Firstly,the spectra of defective and non defective samples of nickel base alloy parts are compared to distinguishing the spectral differences in different samples.Secondly,the spectral base line is corrected by wavelet transform to reduce the blank background of spectral line and deduct continuous background noise.Thirdly,the principal component analysis(PCA)is used to reduce the dimension of LIBS spectral data,which can filter data noise and redundant information and improve the accuracy of defect identification.Finally,machine learning algorithm is used to complete defect recognition,and the recognition performance of the three algorithms is compared and analyzed.The recognition results showed that the LIBS online detection technology combined with machine learning algorithm can effectively identify the defects of SLM metal additive parts.This paper realizes LIBS on-line high-precision detection and identification of metal parts defects in additive manufacturing.Through the self-built laser-induced plasma signal detection and analysis device,the plasma characteristic spectrum has been collected and analyzed.The designed three-dimensional ball screw feed position servo drive system based on LADRC controller can effectively drive LIBS optical fiber detection probe to complete the rapid and accurate positioning of SLM molten pool and samples.The calibration model and classification recognition model established combined with machine learning algorithm not only improve the qualitative analysis performance of different kinds of alloy parts,but also have strong classification and recognition ability for alloy parts defects.The research work and results presented in this paper can provide a basis of defect identification of metal parts made of additive materials,and then provide support for improving the process technology of additive manufacturing,implementing defect mechanism analysis and control,and improving the quality of metal parts made of additive materials.