| In recent years,China’s railway construction is in full swing and plays a decisive role in the development of China’s economy.The catenary wrist arm forms a zigzag frame above the electrified railway to ensure that the train can get continuous current input.Catsnet wrist arm is assembled by seamless steel tube and casing.At present,the assembly and detection of casing on its production line is mainly manual,which has low detection accuracy and long time consuming,which restricts the rapid development of railway construction to a certain extent.By analyzing and sorting out the status quo of wrist arm assembly spatial position detection of catenary,this paper proposes a detection method based on machine vision,optimizes and improves the relevant algorithms of visual image processing,and designs and develops a set of special measurement system for wrist arm assembly spatial position detection relying on Labview and Matlab software.Based on comprehensive and multi-angle experiments,the advantages of high reliability and high accuracy of the system are verified,and the accuracy of the casing assembly position is less than 2～3mm and the casing relative deflection Angle is less than 1～2.The specific research content of this topic is as follows:1.Analyze the current working conditions of wrist arm assembly production and related parameter detection,and study and demonstrate the feasibility of the visual detection system scheme of this subject.In order to adapt to the spatial layout characteristics of the production line,the automatic clamping and rotary positioning mechanism of the wrist arm assembly are designed to realize the automatic clamping and positioning of the wrist arm.Relying on industrial robots to realize automatic loading and unloading of finished wrist arms,an image acquisition system is designed for large size partial and different axis class,supplemented by surface light source to improve the sharpness of the camera’s field of vision,so as to meet the need of image Mosaic pixel registration with detection accuracy less than 3mm.2.Proposed a vision-based method to study the high-precision detection of large-size deviated axis parts.Since the finished product of the wrist arm is of large size and deviated axis,the image collected by a single camera is not enough to capture local detail features.Therefore,the local image of the finished product of the wrist arm is captured by a binocular camera,and the local image of the wrist arm assembly captured by the camera is combined with the improved FREAK algorithm.Compare and analyze the image effect after different filtering algorithm processing fusion,select the optimal processing algorithm.Process images based on mathematical morphology algorithm on the basis of the improved Canny edge detection algorithm to detect bracket assembly contour edge,on edge extraction bracket after finished product contour partially missing,not smooth problem,put forward the least square method based on the minimum deviation probability sum of squares fitting of the extracted contour edge,refining the contour edge pixel features,To achieve the precision of wrist arm assembly space position detection.3.Based on Labview software,a special measurement system for wrist arm assembly space position detection was developed and experimental verification was carried out comprehensively.In order to verify the reliability of the system,the experiment adopted three-dimensional multi-text detection method to capture the wrist arm assembly image from multiple angles.The four casing position parameters and the casing relative deflection Angle parameters of each sample were tested repeatedly for 5 times,and the average value of the error was obtained.The testing results show that the casing assembly position error of each sample is stable within 3mm,the Angle error is less than 2°,and the relative error is less than5%,which proves the advantages of this testing system with good reliability and high precision,and has a good prospect of industrial application. |
PDF Full Text Request