Research On Key Technology Of Labeling Robot For Non-ferrous Metal Casting Production Line

Information identification is a key link in the non-ferrous metal casting production line,and paper labels are one of the main ways for non-ferrous metal information identification.The advantage of paper labels is that the identification information is clear and is not easy to be oxidized and corroded,and it will not cause appearance damage to non-ferrous metals with high requirements.Labels are generally pasted manually.This method is efficient and low-cost,and the on-site operation has certain risks..At present,the common automatic labeling equipment can only move between preset fixed coordinates to complete simple labeling tasks.It is not suitable for non-ferrous metal casting production lines.The automation and intelligence level of labeling equipment needs to be improved.In response to these problems,this paper designs a robot automatic labeling system according to the application requirements.The main research contents are as follows:(1)For the application scenarios of labeling robots,taking lead metal as an example,design the equipment scheme and process of labeling system,and conduct 3D modeling and stress analysis on the key components of the labeling system.(2)Aiming at the image recognition problem of the material position in the labeling process,the deep learning algorithm is used to extract the ROI area and narrow the scope of image processing.In view of the harsh environment and poor lighting conditions in the non-ferrous metal production workshop,the image is enhanced.Improvements are made to solve the shortcomings of the median filtering algorithm,and the filtering effect of the improved algorithm is improved.Aiming at the problem that the linear features of the labeled image are not easy to extract,a linear enhancement scheme of adaptive binarization combined with morphological operations is designed,and the enhanced image linear features are more obvious.(3)Aiming at the problem of detecting the edge straight line of lead ingots in the labeling process,the probabilistic Hough straight line detection algorithm has the disadvantages that the detection angle is too wide and it is easily interfered by continuous discontinuous lines.The probabilistic Hough straight line detection algorithm is improved.In this paper,a Hough space self-attention strategy is designed to find the space where the straight line is most likely to exist,and the detection range of the straight line is concentrated in this space to save computing resources;a global continuous discontinuous line elimination strategy is designed to eliminate continuous discontinuous lines while detecting straight lines to improve the algorithm Anti-interference ability to continuous small and short lines.In this paper,these two strategies are integrated into the original algorithm to obtain an improved algorithm.After experiments,the results are obtained.The average detection rate of the improved algorithm is increased to 1.75 times that of the original algorithm,and the precision rate is increased to1.44 times.Based on the improved algorithm,this paper completes the calculation of labeling points.(4)Aiming at the coordinate system conversion problem of the robot labeling system,the forward kinematics of the robot is solved,and the robot is simulated according to the solution results.Analyze the coordinate transformation principle of the visual system,and establish the mathematical model of coordinate transformation.For the labeling surface alignment problem,perform mathematical operations and adjust the robot pose.(5)Complete the upper computer software design of the robot labeling system based on C# language,deploy the robot labeling system in the factory environment,and complete the robot labeling experiment.Based on the above research work,the results of this paper have been put into practical application in a non-ferrous enterprise in Henan,and the expected effect has been achieved.