Research On Key Technology Of Insulator Surface Damage Detection Based On Image Processing

Insulators are an important component of power transmission lines for aerospace launch sites.In recent years,high-density space launch missions have placed high demands on power systems,but power system accidents caused by poorly damaged insulators have shown multiple trends.The problem of insulator damage detection has become an urgent problem for the power station of the space launch site.For the damage detection method of insulators,it mainly relies on the manual inspection mode,that is,the staff observes through visual inspection.This inspection method faces many drawbacks: the natural environment of Gobitan is bad,the traffic is inconvenient,and the manual method is time-consuming and labor-intensive..In order to solve such problems,many countries have successively launched research on patrol inspection robots for high-voltage transmission lines.Since the mid-1980 s,with the support of the national “863” project,robot inspections have also been greatly developed.The principle is to carry a payload on the robot and build a monitoring and recognition system based on image processing.Combined with the actual space launch site,this paper proposes the design of insulator surface damage detection system based on image processing.The design of the detection system is comprehensively discussed from the aspects of system composition,working principle and function.At the same time,this paper focuses on the key technologies facing the design of detection and identification systems.(1)Study on the image preprocessing algorithm for insulator surface damage.The basic theory and method of partial differential equation applied to noise reduction processing are analyzed.Considering the affine transformation and contrast invariance problem,the AMSS equation is applied in the image preprocessing of insulator surface damage.At the same time,the detailed equation is given.Settlement process.For the key scale parameter selection problems,through the experimental analysis,the method and process of automatically determining the scale parameters are obtained.(2)Research on damage detection and feature extraction method of insulator surface.The key to surface damage detection is to achieve the goal and background segmentation.The reality is that no image segmentation method is universal.Based on the theory of transition area,this paper found that the local segmentation method has a relatively good segmentation effect.Using the information of the transition area,the broken image can be divided into multiple sub-images,and then the image can be segmented by the local threshold.When the number of fixed sub-graphs is eliminated,the effect of the sub-image size on the image segmentation effect is eliminated,which better protects the damaged area.information.After the effective segmentation of the damaged image is completed,three types of features are extracted and described,and the combination is realized by the fusion method,and then the combined feature extraction is performed,so that a more accurate description of the damage can be obtained.(3)For the three typical types of insulator surface damage,this paper proposes a feature description of surface damage by combining features.A method for extracting surface damage features based on kernel principal component analysis and combined feature extraction is presented.Through the application of the kernel principal component analysis method,the correlation between the same eigenvectors at the same surface damage can be effectively reduced.At the same time,the partial least squares method is applied to the damaged composite feature extraction,and different kinds of features under the same damage condition are obtained.The fusion information not only weakens the correlation between different feature information,but also extracts the feature vector to describe the damage feature more accurately.(4)In combination with the damage characteristics of insulator surface,this paper proposes an uncertain SVDD classifier,which achieves the effective fusion of SVDD and damage uncertainty,which is beneficial to improve the accuracy of damage classification identification of insulator surface.According to the uncertainty definition,the SVDD classifier is improved,and the uncertainty support vector machine damage classification method is obtained.Based on the description of the uncertainty of the damaged surface of the insulator surface,the SVDD-based insulator classification method is optimized.Under the condition of limited damage samples,the experimental analysis shows that the combined damage feature can achieve more accurate detection and identification of the surface damage type.