Research On High-precision Vision System For Girder Deflection Measurement Of Gantry Crane

As the main metal structure of gantry crane,the main beam is subject to repeated alternating loads during long-term service,which is prone to fatigue and fracture,causing economic losses and even casualties.Therefore,it is necessary to periodically check its health status.The deflection of the main beam is an important parameter for measuring the health status,and its measurement is of great significance for the safety inspection of the gantry crane.The current methods for measuring the deflection of the main beam have problems such as low accuracy,slow measurement speed,inability to measure dynamically,and high danger.The vision measurement method has the advantages of high accuracy,long distance,and dynamic measurement of multiple positions at the same time.Therefore,this paper studies it and proposes a high-precision vision measurement system for the main beam deflection of gantry crane.Based on the principle of vision deflection measurement,and considering the image characteristics and measurement environment of the main beam of gantry crane,the following research contents are carried out:(1)Design a special light-target and paste it on the bridge girder’s main beam to be measured to enhance the image characteristics of the area to be measured.Because the image characteristics of the main beam are similar,it is not conducive to the recognition and positioning of the point to be measured in the image,and the accuracy of target positioning is easily affected by uneven illumination.Therefore,this paper designs two different shapes of circular and square light-targets.Surface light sources and film masks with special shapes are integrated to enhance the robustness and accuracy of vision deflection measurement.And the positioning accuracy of circular and square light-targets are compared.(2)Study light-target feature extraction and sub-pixel positioning algorithms to improve measurement accuracy without increasing economic cost.According to the grayscale,edge and shape characteristics of the light-target in the acquired image,an improved sub-pixel center positioning algorithm of the light-target is designed.The algorithm is based on the sub-pixel edge detection operator of the grayscale space moment and the RANSAC algorithm.The gravity center positioning method and subpixel template matching method are compared for accuracy.(3)Study the error elimination algorithm caused by camera shake in vision deflection measurement.Due to the influence of the measurement environment,the camera will inevitably produce jitter,which will cause vision measurement errors.Therefore,by decomposing the camera shake into three directions of translation and rotation about three coordinate axes,the principle of error generation is studied and proposed.And a scheme for eliminating camera shake errors using two fixed reference light source targets is presented.(4)Study the deflection calculation algorithm under camera non-parallel photogrammetry.When the camera image plane and the plane to be measured are not parallel,the deflection solution in the physical world and the image world is more complicated.According to the relationship between the target shape parameters and the actual shape parameters in the acquired image,a corresponding deflection solution algorithm is designed.Finally,the simulated deflection measurement test and the deflection measurement test of the main girder of the indoor and outdoor crane under actual working conditions verify the high robustness and high accuracy of the vision deflection measurement system proposed in this paper.The proposed vision deflection measurement system can achieve sub-pixel accuracy.