| Wire and Arc Additive Manufacturing(WAAM)is a typical metal additive manufacturing process.Due to its advantages such as high production efficiency,low manufacturing cost and excellent mechanical properties of forming parts,WAAM has a broad application prospect in the integrated forming of large metal parts.However,since the arc forming process is susceptible to various factors such as process parameters and heat accumulation,the forming precision of WAAM is relatively poor,and the forming precision control has always been the focus and difficulty of research.In this paper,based on the laser vision sensor,a set of arc additive manufacturing forming information detection system is built,and the shape detection and prediction of the weld bead during the arc additive manufacturing process are studied.Firstly,based on the laser vision sensor,a robot wire and arc additive manufacturing and forming information detection system is established in this paper.This system can detect the morphological characteristics of deposition in the process of arc additive forming and predict the morphology of deposition according to the forming parameters.A set of point cloud data preprocessing algorithm is studied by using the point cloud data collected by the visual sensing system to represent the deposition morphology.It includes point cloud data noise reduction based on neighborhood average statistical filtering algorithm,point cloud coordinate matrix rotation correction based on greedy search algorithm,and forming parts section contour obtained based on octree point cloud slice search algorithm.It is proved that point cloud data processing is effective and robust.An algorithm based on the combination of first-order height difference and second-order height difference is designed to select the characteristic points of bead.The quadratic polynomial function is selected to fit the surface profile curve of the bead.Aiming at the detection error of the system,the error was corrected by combining the cross section area of the bead.The prediction model of interlayer temperature,deposition current,deposition voltage,deposition velocity to deposition bead morphology is constructed by using XGBoost algorithm,which solves the problem of less training samples and easy over-fitting of the model in the prediction model.The results show that the average relative error of the prediction algorithm is 1.25% and the prediction accuracy is high.For the multi-layer single-bead WAAM and multi-layer multi-bead WAAM,the accuracy verification experiment of the visual inspection system was carried out.The experimental results show that the average detection error of the visual inspection system is 0.25 mm.Through the aid of the visual inspection system,a series of arc additive manufacturing experiments were carried out,and the resulting molded parts were of high forming quality. |
