| In power operations,safety accidents occur frequently,leading to economic losses and casualties.Then the low-voltage current transformer is an important metering tool in the power grid system.When calibrating the low-voltage current transformer in the metering operation scenario,the calibration device needs to be connected to the low-voltage transformer because the current of the low-voltage current transformer is larger,and accidents are prone to happen in this calibration process.At present,there are some problems in the process of calibrating low-voltage current transformers,such as difficult power outages,time-consuming and labor-intensive disassembly and inspection,and power workers have to wear necessary protective equipment for the operation.In order to solve the above problems,this paper designs and develops a robot arm control algorithm based on machine vision to achieve full-scale calibration of in-operation low-voltage current transformers under energized conditions,and to identify whether the workers are wearing the necessary protective equipment.The combination of machine vision and robotic arm is used to achieve safe access to calibrated current transformers under energized conditions to ensure the safety of personnel,power grid and equipment.The specific work is as follows.(1)For the metering operation site,this paper proposes a YOLOv5-based target detection algorithm and designs a new network structure.The algorithm adopts structures such as no convolutional step or pooling layer,weighted bidirectional feature pyramid network and void space pyramid pooling,which can realize the detection of firewire,zero wire and various markings on low-voltage current transformers.At the same time,the algorithm in this paper can also achieve the detection of the necessary protective equipment for the operator,and once the operator is detected not wearing protective equipment,the system automatically carries out voice alarm.(2)The design of robotic arm vision system based on target detection algorithm,by combining this paper’s target detection algorithm with camera imaging system,through the target detection algorithm to obtain the target location information such as zero wire and fire wire in the low-voltage current transformer.Then the coordinate information of the target is converted into the spatial position information under the coordinates of the robotic arm by calculating through the coordinate conversion and camera calibration of the imaging vision system.(3)The end position information of the end of the robotic arm is known,and by establishing D-H parameter table,positive and negative kinematic analysis and trajectory planning,the trajectory planning algorithm is compared to and the optimized five times polynomial trajectory planning is used to make the end of the robotic arm(actuator)reach the target position smoothly and continuously,and then through the end actuator of the robotic arm,the calibration equipment is connected to the low-voltage current transformer to realize electrical The calibration equipment is then connected to the low-voltage current transformer through the end-effector of the robot arm,enabling the calibration of the transformer under isolated,energized conditions.In this paper,a target detection algorithm based on YOLOv5 is proposed and compared with other target detection models by conducting comparison experiments and ablation experiments on a homemade dataset,and the experimental results show the effectiveness and advancedness of the method in this paper.In addition,this paper also conducts simulation experiments for different algorithms of robotic arm trajectory planning,and the optimized five times polynomial trajectory planning is selected for robotic arm trajectory planning.Finally,the target detection algorithm and the robotic arm are combined to perform the calibration of low-voltage current transformers under electrical conditions by controlling the robotic arm. |
