Research On SLAM Algorithm Of Inspection Robot In Traction Substation

Traction substation is the "heart and blood vessels" and "power source" of the electrified railroad.Substation inspection is the core work to ensure the stable operation of the traction power supply system.Traditional manual inspection has a large uncertainty factor,unskilled business,low efficiency,lack of responsibility and other problems that lead to serious consequences of train downtime.Along with the development of robotics and artificial intelligence,in order to enhance the intelligent operation and maintenance control ability of substation and improve the safe operation and guarantee ability of substation,it is inevitable that the intelligent inspection robot will replace manual and become the main force of substation inspection technology.The inspection robot is positioned in real time according to the a priori map,navigates independently to the inspection point according to the planned path,and collects and uploads data by using the equipped sensors.In this paper,laser SLAM algorithms based on inspection robots are studied to improve the localization,path planning and autonomous navigation performance of inspection robots,as follows:For the problems of ghosting and drift in traditional map construction,a front-end odometer with fusion of LIDAR and IMU data is proposed.Time approximation synchronization is achieved to unify the sensor coordinate system,and point cloud distortion correction is proposed to improve the accuracy of SLAM algorithm by using IMU data to assist LIDAR.NDT point cloud alignment is used to calculate the front-end odometer.Determine the selection strategy of key frames and construct local maps,and experimentally verify that by removing motion distortion,the ghosting and drift problems existing in the built maps are eliminated,and provide high accuracy maps for robot positioning.For the traditional global path planning algorithm is mostly point-to-point,the existing algorithm path planning distance is poor,and the inspection task requires sequential planning and global optimum for multiple inspection points.A dynamic parameter-based improved ant colony algorithm(DP-ACO)is proposed to improve the global path multi-point planning capability.The ideas of dynamic parameter adjustment,elite mediocre ant system and variation are added to address the problem that the ant colony algorithm is prone to fall into local optimality.Finally,the effectiveness of the DP-ACO algorithm is verified by comparison experiments under multiple patrol point conditions with simulation analysis.The DWA dynamic window method is used for the local path planning algorithm,and the evaluation function parameters are analyzed and verified by optimization and simulation experiments.The hardware and software systems of the inspection robot model machine are designed and built,and each module of the system is elaborated.To ensure the accuracy of the platform positioning test,the linear velocity,angular velocity and IMU are calibrated.Finally,experiments of repeated localization under autonomous navigation are designed to verify the effectiveness of the improved path planning algorithm of laser SLAM in this paper.