On-line Detection Of Defects Of Piercing Plug In Cross Rolling Piercing Based On Point Cloud Acquired By Laser Measure And Deep Learning

Hot-rolled seamless steel tube has the advantages of good mechanical properties,easy machining and long service life.It is widely used in the blank,bearing and important pipes of hollow precision parts,examples include precision bearing tubes,pneumatic or hydraulic components,and high-precision structure tubes for vehicles and aircraft.Cross-rolling and piercing is the first process of forming thick-walled seamless steel tube,in which the solid billet is transformed into hollow capillary.In this process,the center of the Steel Bar is pushed through by the plug.It can be seen that the surface profile quality of the plug affects the quality of the inner wall of the steel pipe.In the above process,plug is in high temperature,high pressure,strong friction and rapid cooling hot environment,so the plug surface often appear wear,deformation and other defects.According to statistics,80% of inner wall scratches and thickness inhomogeneity are caused by plug defects,so it is the key to ensure the inner wall accuracy of the tube blank to inspect the plug after each pass of rolling.At present,the main detection methods are life judgment method and artificial vision detection method,which are affected by the composition difference of plug materials and human factors,and have the shortcomings of poor real-time performance,low confidence in detection and poor adaptability to environment,can not meet the requirements of industrial surface defect detection.Therefore,it is urgent to design an intelligent detection system and algorithm on the production line to monitor and analyze the surface defects of the plug after each pass.Based on two-high and three-high cross-rolling piercing mill sets,this paper presents an on-line inspection device and an intelligent algorithm for the plug surface defects of cross-rolling piercing mill,with the financial support of the “Intelligent production line and equipment for serial hot continuous rolling of seamless steel tubes”,which is a major project of Shanxi province,the design and feasibility analysis of the on-line inspection device for seamless steel tube,the on-line acquisition algorithm of the 3D Contour Point Cloud of the plug contour,the locating algorithm of the plug defect,the recognition algorithm of the plug defect and the feasibility analysis are mainly solved.In order to verify the feasibility of the design of the plug on-line detection device,this paper aims at the working cycle,design size,field environment and various defects on the plug surface,in the laboratory,a set of plug on-line detection simulator is designed,which mainly includes a 3D printing model of the defect plug,a plug rotating mechanism,a data acquisition device and a contour scanning device,the feasibility of the design of the plug on-line detection device is verified by the error difference of the scan reconstruction contour of the plug model.In order to verify the feasibility of the intelligent algorithm,the filtering algorithms and applications of all kinds of filters are analyzed at first,the Finite length FIR(Finite Impulse Response)filter can effectively preserve the linear phase of the measured data.Finally,the filtering effect of the filter is analyzed by field experiments.In order to locate the defects on the surface of the plug on-line,a point cloud extraction algorithm for the plug is studied,firstly,the precise registration of Point cloud with standard CAD model is realized by improving the ICP(advanced Iterative Closest Point-RRB-method based on the features of the plug,and after the registration,according to the characteristics of the plug defects in different working areas,the corresponding defect analysis algorithm is applied to classify the defects.