Research And Development On Mounted State Detection And Precise Localization System For Watt-hour Meter Based On Machine Vision

In order to ensure the fairness and justice in electric energy measurement,watt-hour meter should be verified to make sure it can meet the quality and measuring precision requirement.At present,the verification is developing towards automation,standardization,intensification and flexibility.Watt-hour meter needs to be mounted on the verification device during the automatic verification process.And the mounted state detection determines whether the subsequent steps,that is,reliable crimping and accurate verification,can be realized.So the mounted state detection is very important and is the premise of precise verification.This paper researches on the detection of mounted state of watt-hour meter for the extremely large automatic verification system in the electric power research institute of Jiangsu electric power company and on the basis of which studies the precise localization of the incorrectly mounted ones.Firstly,the framework and function of the current extremely large auto-verification system are introduced briefly and a mounted state detecting and precisely positioning system module based on vision is established on its basis.Then the principle and process of the calibration of the vision system are described in detail and the calibration experiments are conducted.Secondly,a method combining machine vision and pattern recognition to detect the mounted state of watt-hour meter is proposed,which can be mainly divided into two steps.The first one is to estimate the perspective transformation matrix between the image of the watt-hour meter to be detected and the templated image created in advanced.The second one is to judge the state using pattern recognition technology and the four types of state can be recognized and fed back timely.Thirdly,a method to precisely locate the incorrectly mounted watt-hour meter is researched.The pixel coordinates of the LED corners are extracted using image process methods,and then the pixel coordinates are transformed into the coordinates in camera coordinate system using triangulation principle.Furthermore,the three-dimensional coordinates in robot coordinate system are reconstructed with the hand-eye matrix and the position relationship between tool coordinate system and robot coordinate system.On this basis,the coordinate system of watt-hour meter is established,which can be used by the robot to pick off the incorrectly mounted watt-hour meter and put it at the right position on the verification device.Finally,the mounted state detecting and precise positioning function module integrated in the extremely large auto-verification system is developed,on which experiments of the mounted state detection and precise localization of watt-hour meter are conducted to verify the validity of the methods proposed.