Parameters Optimization of UAV for Insulator Inspection on Power Transmission Line

In recent years, Unmanned Aerial Vehicle (UAV) technology has been widely used in power transmission line inspection. UAV operating parameters determine the quality of the acquired insulator image, which directly affects the efficiency of transmission line inspection. Regarding the inspection of ceramic insulators, this work establishes a three-dimensional (3D) model of ceramic insulators, and simulates the changes in shooting distance, angle, and height of the UAV through three geometric transformations: rotation, translation, and scaling. The influence of those three UAV operating parameters (the shooting distance, angle, and height) on the integrity of the ce-ramic insulator steel cap is studied. In the infrared zero value detection of 110kV ceramic insulators, It is found that the best setting for obtaining the maximum steel cap area is: shooting distance of 3–7 meters, shooting height of 1 meter below the center of the insulator string, and a shooting depression angle of 0°. The UAV operating parameters setting method proposed in this paper will provide theoretical and practical basis for the in-depth intelligent inspection of transmission line insulators.

ping, and deformed) insufficient will cause problems such as 23 low robustness on defect detection [3]- [5], resulting in poor 24 detection results. 25 Several researchers have studied insulator identification 26 and defect detection based on insulator images. Jiang et al. [6] 27 proposed a method to identify insulators by combining the 28 The associate editor coordinating the review of this manuscript and approving it for publication was Zheng H. Zhu . color, shape and texture characteristics of insulators, and 29 proposed a method to detect the defects of glass insulators 30 based on the distance between the insulator plates. However, 31 if the shooting distance is far, the influence of the background 32 texture will increase, resulting in a decrease in the detec-33 tion accuracy. Zhang et al. [7] identified insulator defects 34 based on particle swarm optimization feedforward neural 35 network He et al. [3] used the Faster R-CNN algorithm [8] 36 to build an insulator detection model and locate the insulator 37 target. Then, he realized the lack of defect discrimination 38 of glass insulators through CNN algorithm. But when the 39 adjacent insulators are blocked, the detection result was poor. 40 Zhou [9] and others proposed a degraded insulator iden-41 tification method based on the combination of time series 42 model and infrared detection technology, but the steel cap 43 area need to be separated from the infrared image. When the 44 insulator steel cap is blocked by the adjacent insulator, it will 45 decrease the detection accuracy. Therefore, by optimizing 46 the operating parameters of the UAV and standardizing the 47 drone.The equivalent correspondence between the changes of 83 UAV operating parameters and the geometric transformation 84 of insulators is shown in Figure 1. Figure 1(c) shows the corresponding relationship between 100 the shooting angle and the scaling and rotation of the insu-101 lator. The shooting depression angle can be equivalent to 102 rotating and scaling the insulator. The rotation method is 103 by rotating the insulator string in the vertical direction with 104 the UAV as the base point, and the rotation angle is equal 105 to the shooting depression angle; the functional relationship 106 between the shooting depression angle and the zoom ratio r 2 107 is shown in equation (2).
By analyzing the corresponding relationship between the 110 shooting height, distance, depression angle and the geometric 111 transformation of the insulator, according to the superposi-112 tion principle, the height, distance and angle of the insulator 113 shooting by the UAV can be simulated.

116
The insulator used in this article is the XP-160 disc type 117 suspension ceramic insulator. The cross-sectional view is 118 VOLUME 10, 2022   Insufficient image completeness of the insulator will cause 151 the lack of key information in the image, which is one of 152 the reasons for missed detection. The insulator area ratio is 153 defined as the ratio of the insulator area to the reference 154 area value. The insulator area ratio is used to quantify the 155 integrity of the insulator image, where the area reference 156 value is the total area of the insulator under unobstructed 157 conditions. The closer this insulator area ratio is to 1, the more 158 complete the insulator area in the image, the less overlap and 159 occlusion.

160
For different types of insulator defects, the characteristic 161 areas of concern are different, and appropriate evaluation 162 indicators need to be determined according to the type of 163 defect. The early decay of composite insulators is often man-164 ifested as abnormal heating of the sheath [18]- [22], and the 165 integrity of the sheath can be used to quantitatively evaluate 166 the infrared inspection image of the composite insulator. 167 The zero-value insulator is mainly manifested as abnormal 168 heating in the steel cap area [14], [23], and the inspection 169 image of the zero-value insulator is based on the integrity 170 of the steel cap. Taking zero-value insulator detection as an 171 example, the calculation method of steel cap area ratio is as 172 follows.  (3).  and an optimal plan for operating parameters selection was 240 proposed.

241
The variation law of steel cap area ratio with shoot-242 ing depression angle and shooting height is shown in 243 Figures 6 and 7. The x axis is the shooting height, and the 244 VOLUME 10, 2022 y axis is the steel cap area ratio. When the depression angle 245 is greater than or equal to 0 • , the area ratio of the steel cap  ratio of the insulator steel cap, the shooting distance should 281 be 3-7m. The constraint condition of the shooting distance 282 d is expressed in equation (4). The structure and shape of the 283 insulator determine the value of the threshold angle. When the 284 shooting depression angle is greater than the threshold angle, 285 the shooting height should be taken as the lowest height for 286 obtaining the entire insulator image, as shown in equation (5). 287 When the shooting depression angle is less than the threshold 288 angle, the shooting height should be the highest height to 289 capture the entire insulator, as expressed in equation (6).
This paper has discussed the optimization of inspection 294 parameters for vertical pendant insulators. When facing a 295 porcelain insulator in a horizontal or inclined state, the 296    area ratio and the shooting height is parabolic, and it gradually 334 transforms into a monotonically increasing curve after the 335 elevation angle continues to increase, which is consistent with 336 the simulation results obtained in Section 5.2. 337 VOLUME 10,2022 According to the simulation results obtained in Section 5.2 and the field experiment results, the variation of the maximum 339 steel cap area ratio with the shooting depression angle is 340 compared, as shown in Figure 12. It can be found from the 341 figure that the area ratio of the steel cap in the field experiment 342 is slightly lower than the area ratio of the simulated steel 343 cap, but the absolute error is within 0.14. The reason may  3) The shooting distance should be within 3 m∼7 m.