Research On The Detection Method Of Traction Wheel Wear Based On Machine Vision

The traction elevator is widely used in people’s daily life.The traction wheel provides power for the traction machine through the friction between the internal rope groove and the traction rope.In the process of reusing,the traction wheel will wear gradually,which will lead to the failure of the traction system and bring hidden danger to the safe operation of the elevator.At present,the detection of traction wheel wear is mainly based on qualitative analysis,which can not achieve quantitative and accurate detection.This paper studies the characteristics of the rope groove structure of the traction wheel,and studies the non-contact measurement method of the traction wheel groove wear through the machine vision technology.This thesis first investigates a large number of common groove types of elevator traction wheel,summarizes two failure forms,and summarizes the quantitative detection standard of traction wheel wear from two aspects of national standards and daily inspection and maintenance experience of elevator traction wheel.Based on the analysis of the physical structure of the traction wheel groove,a traction wheel wear detection scheme based on machine vision is constructed,and a non-contact quantitative traction wheel rope groove wear detection method is proposed.According to the actual detection environment and the characteristics of the tested object,the hardware selection and design of the traction wheel vision detection system are completed,and the software work flow is analyzed in detail.The five-slot traction wheel is taken as the test object.In order to reduce the influence of camera itself and external factors on the image quality as much as possible,the source image is enhanced,the imaging noise is removed by frequency domain filtering,and then the edge is enhanced by Laplace operator to highlight the edge information in weak light environment,To avoid the subsequent target slot matching and feature point detection error.In the case of fuzzy imaging and uneven illumination,it still has high accuracy.The image of the traction wheel groove obtained by the camera is greatly affected by the illumination.Due to the complex texture structure of the traction wheel groove,there will be uneven illumination,many dark spots,and the edge transition is not obvious.This paper demonstrates that the environment of the traction wheel is not suitable for morphological contour processing by analyzing the intensity of the illumination mode,and finally adopts the method of feature point detection to highlight the contour boundary.Through the construction of wire rope texture features and weighted fusion,the redundant feature points are removed and only the wheel groove boundary feature points are retained.Through the curve fitting analysis of boundary feature points,the inflection point of wire rope is calculated,and the distance between rope grooves is measured.In this thesis,the physical model is built based on the system scheme,and the error between the calculated value and the actual result is discussed from three aspects: the axial distance of the camera,the distance between the camera and the rope groove,and the shielding error of the traction wheel groove,and the control and compensation are carried out.In this paper,the high-precision feeler gauge is used to take the mean value of multiple measurements as the standard value.Finally,the distance between the rope grooves is measured by the detection algorithm in this paper.The measurement results are compared with the standard value to verify the feasibility and experimental accuracy of the system.