Vision Technology Research And Application For Robot Operations In Distribution Station And Substation

With the rapid development of power system in China,the demand for operation and maintenance of power system is increasing with the continuous expansion of power system scale.The traditional operation and maintenance based on human model is gradually difficult to meet the growing requirements of power system operation and maintenance.State Grid Corporation is actively promoting the establishment of intelligent operation and maintenance system and the innovation of operation and maintenance mode of power grid.It is the research and development direction of intelligent operation and maintenance of power grid to use advanced intelligent robots instead of manual operation,which has been put into practical use in some electric power fields.In this paper,the key technology of distribution station and substation operation robots,machine vision technology,is studied based on the project of "Intelligent Substation Isolation Circuit Breaker Live Maintenance Robot" and "Multi-Task Cooperative Distribution Station Duty Robot" cooperated with State Grid Corporation.Aiming at the differences of indoor and outdoor operation target and environment,the highly robust visual positioning and recognition methods are designed.The effectiveness of methods are proved by a series of experimental tests and applications.The main contents of this paper are as follows:1)The overall scheme design of vision system for distribution station and substation operation robots is presented.According to the visual task requirements of distribution station and substation operation robots,the overall framework of vision system is constructed.According to the different designs and application scenarios of the robots,monocular vision platform is built for the live maintenance robot of substation and binocular vision platform is built for the on-duty operation robot of distribution station.2)The monocular and binocular vision positioning methods are designed.Aiming at the analysis of the principle of camera imaging and three-dimensional positioning in hand-eye system composed of camera and manipulator,the calibration analysis of hand-eye vision system is carried out.According to the difference of image information obtained by monocular and binocular vision,a method of locating image information by identifying features with fixed constraints is designed for monocular vision,and a method of locating image information by fitting two images is designed for binocular vision.3)The features that meet the requirements of high precision positioning are designed and selected.Because the isolation circuit breaker of substation does not have the characteristics of high precision positioning,by analyzing the advantages and disadvantages of various positioning features,a threeMark circle feature with fixed size is designed for precise positioning.For the three switches of the distribution cabinet,the rectangular white panel of the remote on-off switch and the closing-opening switch,and the circular face frame of the rocker arm cart are selected as the positioning features of high accuracy.4)The image processing algorithm for indoor and outdoor complex environmental impacts is designed.For complex indoor and outdoor imaging environment,image processing algorithm designs an adaptive threshold segmentation method based on brightness optimization and combines closed operation,median filtering and Canny operator edge extraction to eliminate the influence of brightness,noise,imaging quality and other factors,so as to accurately extract the contour edges of location features from images.The least squares ellipse fitting method and the least circumscribed rectangle fitting method are designed to fit the edge to get the geometric information of the location feature image.5)Switch type and state recognition algorithms are designed.According to switch type recognition,the difference of switch characteristics is analyzed,and a template matching algorithm based on gray features is designed and its adaptability to brightness change and interference is optimized.According to the profile characteristics of the switch knob,an ellipse fitting method based on boundary distance optimization is designed to reflect the deflection of the switch knob and judge the state.6)Experimental testing and development applications are implemented for verification.Based on indoor and outdoor environments,a series of experiments are carried out to test the positioning and recognition ability of the vision system,which proves the robustness and effectiveness of the vision system.Through the cooperative operation tests between vision system and distribution station and substation robots,it is verified that the visual system can complete the visual tasks of the distribution station and substation operation robots,and assist the robots to complete the operations accurately,which proves its practical value.