Research On Geometric Parameter Measurement Technology Of Rigid Contact Network Based On Binocular Line Array Vision

With the rapid development of rail transportation,how to ensure the safety of the transportation system has become the focus of the railroad sector.As the main channel of power transmission in electrified railroad,the geometric parameters of contact network directly affect the safety of pantograph power supply.However,the current contact network geometry measurement mainly adopts the method of structured light-assisted visual measurement,which is simple but limited by the frame rate of the surface array camera and can hardly meet the requirements of high-speed measurement on board.Therefore,how to improve the detection frame rate while ensuring the measurement accuracy is the main problem that needs to be solved at present.Considering the advantages of line array vision in high-speed imaging and measurement,this thesis adopts the idea of binocular vision and investigates the technology related to the measurement of contact network geometric parameters by binocular line array vision,and the main work includes the following aspects.Firstly,a complete set of static calibration method based on binocular line array is proposed to address the problem of limited calibration environment caused by the mainstream dynamic scanning imaging method in the current line array camera calibration method,which needs to maintain uniform motion between the precision displacement platform and the calibrator.The method is based on the transformation relationship of the visual coordinate system of the line array camera to obtain the minimum number of calibrators and their characteristics required for the focal length calibration of the line array camera,and a reasonable and scientific calibration plate scheme is designed.In addition,in order to improve the accuracy and flexibility of the binocular line array measurement model,this thesis proposes a calibration method for the relative position of binocular line array based on the binocular line array measurement model,combined with the imaging characteristics of the line array camera.The experimental results show that the measurement accuracy based on this method can reach0.50 mm under ideal laboratory conditions,with an error ratio of 4.5%,which provides a theoretical basis for the subsequent geometric parameter measurement of the contact network.Secondly,for the problem of poor localization in the measurement of geometric parameters of rigid contact networks,an improved TLD localization algorithm is proposed based on the advantages of the TLD algorithm in the long-time tracking capability and the problems in the localization of rigid contact networks.The algorithm is improved on the basis of the traditional TLD algorithm,which is replaced by an improved kernel correlation filtering algorithm SVM in the detection phase,so as to increase the detection accuracy under the change of light intensity.In the tracking stage,the motion characteristics of the contact network are introduced and the improved KCF algorithm is used to optimize the detection module,so that the algorithm can better solve the problems that the traditional TLD algorithm cannot solve the transient movement of the target position and cannot identify the "double line".The experimental results show that the algorithm can effectively improve the positioning effect of the contact network.Based on this positioning effect,the techniques related to the geometric parameter measurement of the contact network are studied,mainly including edge detection and geometric parameter calculation.Combined with the measurement model and calibration method proposed in Chapter 3,the experimental results show that the geometric parameter measurement method of the contact network in this thesis has a high accuracy through the self-made data set for validation.Finally,in order to verify the effectiveness of the method proposed in this thesis,a contact network geometric parameter measurement system is designed and implemented.The model is deployed in the upper computer,and the operation results of the prototype system are demonstrated by collecting simulated contact network data.The example shows that this prototype system can effectively accomplish the measurement function of contact network geometric parameters.