Research On Laser 3D Printing Molten Pool Monitoring Technology Based On Machine Vision

Laser metal deposition(LMD)is a typical metal additive manufacturing technology,which can be used to manufacture large,high-performance and complex components through dynamic control of process parameters for aerospace,rapid prototyping,biomedical and other industries.The physical and chemical processes involved in the LDED process are complex,including laser attenuation,heat transfer,laser interaction with powders and molten metals,molten pool dynamics,and solidification behavior of the molten pool.So,it is difficult to control process conditions and to predict component performance.Among these process parameters,the molten pool is the most basic unit.Its size and shape during the process are significantly related with the process parameters and have important impact on the mechanical properties of the components.Many technical methods have been adopted to obtain abundant molten pool morphology information.This information can be used as the basis for performance prediction and closed-loop control.However,the signals obtained were not sufficiently applied and properties of the components were not satisfactorily predicted.In this study,the superalloy IN625 was selected as experimental material for the LDED process.Distribution of the size and morphology of the molten pool were extracted by a machine vision technology.The relationship between the distribution of the size/morphology,the process parameters and the final microstructure/properties of the component was investigated.The results obtained in this study will provide a reference for process control and component performance prediction.The main tasks are divided into the following aspects:(1)Open CV was used to process molten pool images and extract feature values.The area and perimeter of the molten pool were obtained,and the compactness of the molten pool area was further calculated.Using area and compactness as the two main molten pool characteristic values,multi-tracks multi-layers LMD experiments under different line energy densities were conducted.The two molten pool characteristics were analyzed in the time domain.As the process progresses,the both characteristic values fluctuate around a basic value.Different instability phenomena exhibit different fluctuation frequencies and amplitudes,which can be used as effective monitoring data in the process.In addition,the typical molten pool images of unstable phenomena involved in this article and the causes of unstable phenomena are discussed in detail.(2)Taking all the layer heights,closeness to the edges,and linear energy density measured at each layer,the mean and standard deviation of the molten pool characteristic values were calculated.Changes of the two statistic characteristic values,and the state of the process system represented by the data changes were analyzed.The results show that the mean area of the molten pool is more sensitive to the determination of powder feeding stability,while the mean and standard deviation of the molten pool compactness are sensitive to the instability of the laser during the initial buffering process.Moreover,considering the influence of the linear energy density on the surface quality of the sample block,the process stability control conditions for improving the surface quality may be represented by the molten pool area.(3)An optical microscope was used to analyze the morphological characteristics of the sample manufactured under different layer heights,sampling positions,powder feeding stability,and linear energy density.The evaluation coefficients of the arrangement regularity of the molten pool were designed in order to quantify influence of the different linear energy densities and sampling locations.Forming quality,tensile properties and Vickers hardness of the samples were studied.The results show that when the linear energy density is very low,the property consistency of each position of the sample block is poor.The characteristics of molten pool eigenvalues in time domain,the variation of sample statistical values with respect to various process conditions and the discriminability of different types of instability were discussed.Finally,the relationship between process,molten pool,microstructure and properties was discussed under different linear energy density.In this study,multi-tracks multi-layers characteristic values of the molten pool were used to quickly identify unstable phenomena,enriching the data base of LMD process,optimizing rapid process parameter selection and early detecting defects.The results,therefore can promote development of LMD process in industrial applications.