Research On The Intelligen Detection Technology Of Porcelain Deteriorated Insulator Based On Infrared Thermal Image

As an important power equipment,insulator is widely and largely used in power system,which has great influence on the secure and stable operation of power grid.The porcelain insulator work long hours in the harsh condition with high electric field,large mechanical stress,serious pollution and complex humiture,coupled with the defects of its own material properties.In this case,the insulation performance of the insulator may easily be faulty and lead to power grid security accidents.Therefore,it is quite necessary to detect the state of insulator.The current deteriorated insulator detection method has many problems.The infrared thermal image method has incomparable advantages in insulator state detection field.This paper investigate the intelligent detection technology of porcelain deteriorated insulator based on infrared thermal image.The main contents include the following aspects:1.Systematic and detailed analysis in the principle of heating theory and the infrared detection technology of porcelain Insulator.On the basis of theoretical research,a series of indoor simulation experiment were carried out:The experiment about the influence of environmental factors on the insulator string's infrared characteristics;The experiment to obtain the insulator string's infrared characteristics when faulty insulators in different position;The experiment to obtain the insulator string's infrared characteristics when the faulty insulator's resistance changes.2.Proposed an automatic extraction method of the insulator string's steel caps and disks area in infrared image.Through gray processing,bilateral filtering,OTSU binary segmentation,make infrared image preprocessing;Using the specific algorithm to realize the automatic location of insulator string;Using the linear regression of the insulator string's skeleton to correct the image angle;According to the distribution of insulator string's columns width,segment insulator string and extract insulator areas;Using the Fourier Descriptor for insulator areas identification;Separate the steel cap and disk,realize the extraction of the steel caps and disks area.3.Put forward a faulty insulator recognition model,which is based on the improved online sequence extreme learning machine.Using the particle swarm optimization algorithm,realize the automatic search of OS-ELM optimal network structure;According to the result of image processing,automatic extract the insulator infrared characteristic information from the infrared image;Using the recognition model to process the information.The experimental results demonstrate the correctness of theoretical analysis,providing direct support to the investigation of infrared intelligent detection technology.It is observed that the steel cap and disk area of the insulator can be accurately extract in complex background by the image processing method proposed in this paper.Moreover,the faulty insulator can be identified more accurately and quickly by the recognition model used in this paper.