Research On The Method Of Zero Value Insulator Identification Based On Infrared Imaging

The Insulator is an important component and a huge number of equipment of the overhead transmission line. Overhead line insulators work in the outdoor environment, they are subjected to the long-term erosion of environmental, mechanical and electrical load, conductor galloping, etc. They will become zero value insulators because of their mechanical properties and insulation properties reducing. The existence of the zero value insulators is the hidden danger of the safe operation of power system, it will make the whole string insulator voltage distribution distortion and an overall decline of the insulation performance. Under the bad weather conditions, it will make the line flashover greatly increase, even lead to serious accidents like the insulator string off or wire short. The detection of zero value insulators is an important work of the electricity sector, but also the overhaul of the transmission line state. Currently, the methods for zero value detection have been used in field testing applications at home and abroad, and they provide effective means for the detection of zero value insulators, but they also have some disadvantages, such as heavy workload, long cycle, the high cost, poor security, low accuracy etc. With the development of infrared imaging technology, it has been widely used in transmission and distribution equipment fault detection because of its accuracy and efficiency. It is great significant to put the infrared imaging technology into the detection of zero insulators, the level of safe operation of the grid will be improved.In this paper, a systematic study of the identification method for zero value insulators has been made based on infrared thermal imaging technology. On the basis of theoretical studies and experimental verification, the method of using the relative temperature distribution of the insulator string and the artificial neural network to identify zero insulators is proposed, it can be used for zero insulator identification in different contamination level and humidity conditions. Get the infrared images of suspension insulators in simulation110kV line experiment. Because the noise types and multi-level gray-scale effects characteristics of the infrared image, the selection of the adaptive median filter and neighborhood averaging method are combined for infrared image denoising, and then on this basis, the D maximum entropy threshold segmentation algorithm is selected and the infrared image and background areas in the insulator strings are segmentated effectivly. Becaues of the good effects of image pre-processing, image noise is effectively suppressed and the quality of the image is improved; the target area is extracted with the clear outline. Based on the thermal degradation and distribution characteristic of the the zero value insulators and the outside environmental conditions, primary characteristic parameters are analysised and identified by one-way ANOVA statistical analysis, and the essential characteristics which can accurately reflect to significance infrared image of the zero value insulators are selected. The input parameters of the recognition model can effectively determined. Combined with relative humidity, tempratureand and equivalent salt deposit density as the input vectors of identification model, and the actual state classification information of whether the insulator string with zero insulators as the output vectors, the optimized identification model trained is used to identify zero insulators based on improved BP neural network, and the model is tested through the experiment data. The result of test indicates that this method is accurate and effective, it can provide reference for the maintenance work of the power transmission line insulation condition.