Research On Hyperspectral Detection Method Of Pollution Types And Salt Deposit Density Of Transmission-line Insulators

Insulators are used in a huge amount in the power grid system and play an important role in ensuring the safety of transmission lines.The equipment needs to work in the atmospheric environment for a long time,so dust particles in the air are easy to accumulate on the surface,and pollution flashover is easy to occur in weather conditions with high air humidity.At the same time,UHVDC transmission projects are also developing vigorously.Insulator fouling is much more serious in DC electric field than in AC electric field.Pollution flashover is a very complex problem,which is affected by many factors,among which pollution composition and pollution amount are the main factors.The type of typical pollution components determines the pollution type of the insulator,and the amount of pollution determines the pollution degree of the insulator.In this paper,the pollution type and pollution degree of insulators are used as detection targets,and hyperspectral spectroscopy is used as a technical means to realize multiangle and non-contact detection of pollution status of insulators.Taking silicone rubber sheet as the pollution research object,samples of different pollution types and degrees were prepared with typical pollution.The hyperspectral images of all samples are collected and preprocessed to remove noise and other interference information,and the spectral line data of samples are extracted and analyzed;The insulator pollution type and pollution degree detection models based on support vector machine algorithm are established respectively,and the joint detection model of pollution type and pollution degree is further established,which can further grasp the pollution degree when the insulator pollution type is known.Furthermore,the detection methods of on-site insulator pollution type and pollution degree are studied.The salt density of field insulators was measured in different regions.After collecting the hyperspectral images of insulators,the spectral line data under different pollution levels were extracted,and the characteristic bands of insulator hyperspectral were obtained by continuous projection algorithm;Based on the partial least squares regression algorithm,the model between reflectivity and salt density at the characteristic band is established,and the method of salt density spectral density is proposed to calculate the statistical characteristics of insulator salt density,so as to realize the accurate detection of salt density value of all pixels on insulator disk;The whole salt density method is also proposed to evaluate the pollution degree of the whole insulator,and finally realize the visualization of the pollution of the insulator disk.The method proposed in this paper to detect the pollution type and pollution degree of insulators can reflect the pollution information and pollution accumulation state of insulators from multiple angles,and provide a new means and method for the detection of insulator pollution state.