Design And Implementation Of Target Defect Recognition Module For Inspection Robot

As an important part of power transmission,substations are especially important for their safety management.In recent years,the intelligent inspection of substations has continued to advance,and the use of inspection robots in substations has become more and more widespread.However,with the continuous deepening of the use of substation inspection robots,substation inspection robots also face many new problems that need to be solved.Among them,the oil leakage defects of a large number of oil-filled equipment in substations seriously endanger the safety of substation operations.It is very important to identify the oil leakage defects of substation equipment timely and accurately.Therefore,new requirements are put forward for the target defect recognition function of the substation inspection robot,but the current substation inspection robot does not have a module specifically for the identification of substation equipment oil leakage defects.Therefore,this paper applies deep learning technology to the identification of oil leakage defects of substation inspection robots,and gives a recognition network specifically for oil leakage defects of substation equipment.At the same time,a software and hardware environment for oil leakage identification of substation equipment is built.Visualize the software interface of the operation and identification function,and compress and transplant the oil leakage defect identification network into the environment,and finally obtain an independent oil leakage defect identification module of the substation inspection robot.The main research work of this article is summarized as follows:1.Demand analysis and data set construction of oil leakage recognition module for substation inspection robot.This paper summarizes and analyzes the current status of target defect identification of substation inspection robots and the unresolved identification problems.At the same time,it analyzes the current urgently needed oil leakage recognition function requirements of substation inspection robots.Aiming at the target defect recognition of substation inspection robots,the construction of deep learning framework is introduced in detail,and the oil leakage recognition data set of substation equipment is constructed,which lays the foundation for the research of the inspection robot substation equipment oil leakage recognition method.2.Improvement of the identification method of the oil leakage defects of the inspection robot substation equipment.Based on the analysis of substation equipment oil leakage identification method and deep learning network,combined with the characteristics of substation equipment oil leakage defects,a neural network suitable for inspection robot substation equipment oil leakage defect identification was selected and improved.Finally,a test experiment was carried out on the improved network.The network mainly draws on the advantages of ResNet and VGGNet.By eliminating the downsampling operation and replacing the convolution kernel,the improved method effectively reduces the loss of original image information,especially the information loss of small objects.Experiments show that the improved network structure effectively improves the accuracy of the substation equipment oil leakage defect identification network,especially for small targets.3.Design and implementation of oil leakage recognition module for substation inspection robot.In response to the current needs of the substation inspection robots for the function module of oil leakage defects of the equipment,by building a software and hardware environment for substation equipment oil leakage identification and writing a software interface with visual operation and identification functions,the equipment oil leakage defect identification network is compressed and transplanted In this environment,the oil leakage defect recognition module of the substation inspection robot was obtained and tested.The experimental results show that the module has good performance,which can not only provide the robot with information on oil leakage defects in the substation,but also meet the customer’s use requirements.