Research On Path Planning And Target Recognition In Fault Inspection Of Flying Robot For Pylon

Pylon inspection is a necessary link to ensure the normal operation of the power grid.In the research of automatic patrol inspection of pylon,how to ensure the safety of workers in inspection and improve the efficiency of inspection is a very important hot direction.In recent years,with the progress of UAV's flight control technology,sensor technology and AI technology,flying robots equipped with various sensors and intelligent technologies have been applied to many fields.The flying robot can replace the manual to complete the inspection task to ensure the safety of personnel and inspection efficiency is greatly improved,therefore,it was gradually applied to the inspection of power transmission tower and transmission line.In the process of realizing automatic fault inspection of electric tower,the path planning and target recognition of tower are two key components.This article has carried on the corresponding research to these two parts,carried out the following work:In the first part,a 3D route planning method for UAV Based on Improved Grey Wolf optimization algorithm is proposed for the path planning problem of UAV in 3D complex environment.In order to simulate the real geographical environment,a three dimensional geographic model and a no fly zone model are established,and a reasonable track evaluation model is constructed.In the improved algorithm,aiming at the low individual fitness in the initialization process of standard GWO,a population initialization method based on greedy thinking and mutation strategy is designed,which improves the fitness degree of individuals.Aiming at the problem of poor search ability of standard GWO algorithm,a nonlinear function is introduced to distance control parameters,which improves the search ability of the algorithm and avoids the local minimum to a certain extent.Aiming at the problem that the standard GWO algorithm is not flexible in location update,a dynamic weighted average and static average mixed update location policy is designed,which improves the flexibility of location update policy.Finally,the experimental results show that the algorithm has less track cost and faster convergence than many related algorithms.The second part,aiming at the problem of target recognition of electric pylon and the problem of too harsh shooting and incomplete inspection,a method of tower target recognition based on point cloud registration and blocking idea is proposed.The corresponding semantic label model and security model are established for the pylon.In the improved algorithm,the pylon is partitioned,then each piece is roughly matched and finely matched.The rough matching adopts the registration method based on the PCA algorithm,and the fine matching adopts the registration method based on the ICP algorithm.Finally,the experimental results show that,without setting too many preconditions,this method can identify the whole pylon accurately and conveniently.