Research On Automatic Detection And Classification Of Defects Of SLM Powder Bed Based On Machine Vision

Selective laser melting(SLM)technology has the characteristics of short manufacturing cycle,free from the restriction of complex structure,high precision and low surface roughness.It is widely used in the manufacturing of aerospace parts,and the reliability of SLM forming parts is the focus of current research.SLM parts are formed by molten powder,and the uneven powder bed will affect the quality of the parts.Therefore,it is very important to identify the defects of powder bed and control the uniformity of powder bed.The existing manual identification methods have some problems,such as easy to miss and mischeck,low efficiency,unable to obtain defect related data and record them.In this paper,the automatic detection and classification of powder defects in laser selective melting powder bed based on machine vision is carried out,and the defect detection,classification and quantitative characterization of powder image are realized,which is convenient for subsequent integration into the forming equipment and to realize the real-time monitoring and automatic processing of powder defects.Firstly,the defect detection system of SLM powder bed based on machine vision is designed and built.Based on the analysis of the causes of powder defects and the actual needs,the detection system is designed,and the appropriate industrial cameras,lenses and other hardware equipment are selected to build the system.The industrial camera is installed in the way of side axis,the camera is calibrated and the angle of view is corrected.Secondly,an automatic detection and recognition algorithm for different defects of SLM powder coating image is proposed.It mainly includes image segmentation of low gray defects based on probability fast automatic threshold selection,low gray block defects extraction based on morphology and connected domain noise removal method,low gray line defects extraction based on improved Hough transform algorithm and high gray defects extraction based on lag threshold image segmentation and denoising method Different kinds of defects are extracted to achieve preliminary classification.Finally,according to the defect characteristics of SLM powder bed image,the defect quantitative characterization index with strong discrimination is analyzed and designed,and the defect classification model of SLM powder bed image based on CART binary decision tree algorithm is constructed.When detecting and classifying the powder spreading defects,the corresponding index values of defects can be output and recorded at the same time.The corresponding experiments are designed to verify the automatic detection and classification algorithm.The results show that the classification accuracy of SLM powder bed powder image defects can reach 96.67%,and the detection time of single image is less than 0.5s.