Research On Inspection Technology Of Substation Robot Based On Multi-Source Heterogeneous Vision

Substation is the cornerstone and hub of the power system,and its health status plays a vital role in the safe operation of the power grid system,and has a significant impact on the development of the national economy and the safety of people’s lives and properties.Therefore,the status monitoring and fault diagnosis of substation equipment has important practical significance.Due to the wide variety of substation equipment,wide distribution and large environmental differences,manual inspections are currently used to monitor and diagnose the working conditions of substation equipment.However,it is affected by various factors such as labor intensity,business level,and weather environment.The method is prone to problems such as missed inspection,false inspection.With the development of robot and visual inspection technology,the inspection technology that combines the two organically can replace manual methods,has obvious technical advantages and become one of the research hotspots.However,the robot moves to the predetermined position and stops to adjust the PTZ and camera parameters to complete the basic image acquisition,which severely restricts the inspection efficiency.In addition,it is difficult for single-source vision to comprehensively describe the information of different characteristics such as the structural size,shape composition and temperature distribution of substation equipment,and the existing visual detection and recognition methods implemented by deep learning have problems such as limited performance and weak versatility.Based on the above background,it is proposed to carry out research on substation robot inspection technology based on multi-source heterogeneous vision.On the basis of understanding the research status and development trends of substation robot inspection related technologies,determine the overall technical plan for substation robot inspection,focusing on key technologies such as integrated motion control of robots and PTZ,multi-source heterogeneous vision detection and target recognition.Complete the research and realize the integration of the system.Related experiments are carried out to verify the technologies studied in this paper.This paper presents the following contributions:In chapter 1,I discussed the significance of substation operation and maintenance and the research on inspection technology,summarized the research status and development trend of the related technology of substation robot inspection,and analyzed the problems and corresponding problems in the current substation robot inspection technology.The countermeasures clarified the research direction of the thesis.At the same time,the research content and chapters of the thesis were arranged.In chapter 2,I clarified the technical framework of multi-source heterogeneous machine vision,analyzed the imaging mechanism of visible light and infrared and its influencing factors,introduced digital image processing and image analysis methods based on deep learning,and established the theoretical basis of multi-source heterogeneous machine vision.At the same time,the functions and performance goals of the robot system for substation inspection were clarified,the design of the overall plan of the inspection system was completed,and the key technologies that needed to be solved were condensed.In chapter 3,the research on the integrated motion control technology of the inspection robot and the pan-tilt was carried out.On the basis of clear three-dimensional space pose description method,the internal and external parameter modeling and parameter calibration of the camera were completed,the kinematics analysis of the robot and the gimbal was carried out,the integrated kinematics model of the robot and the gimbal was established,and the dynamic image data was realized.Acquired visual servo control.In chapter 4,I carried out the research of multi-source heterogeneous machine vision detection and target recognition technology.Using visible light and infrared cameras to collect various equipment status images of the substation,the production and expansion of the image data set was completed.At the same time,YOLOv3 was used to construct a multi-source heterogeneous visual generalized detection model,and integrated the path aggregation network and deep separable convolution,which not only enhances the feature extraction ability,but also reduces the scale of the network model and improves the real-time detection.So as to achieve high-performance multi-source heterogeneous vision detection and target recognition.In chapter 5,based on the theories and technical achievements described in the above chapters,adopted distributed architecture and modular design,organically integrated multi-source heterogeneous visual inspection and mobile robot technology,and the development of key software and hardware modules is completed.Carried out the system integration,and developed a set of substation robot inspection system.At the same time,the system is used to carry out related inspection experiments such as motion control and visual inspection to verify the feasibility and effectiveness of the technology and system developed in this paper.In chapter 6,the research content and innovations of this paper are summarized.Make a prospect for further research content and direction.