Research Of Insulator Fault Detection System In Unmanned Substation Based On Machine Vision

With the improvement of power grid dispatching automation, the operation of unmanned substation becomes the developing trend in current power system. Remote video monitoring is used on the basis of the traditional“four-remotion”(remote measurement, remote communication,remote control, remote adjustment). Remote video monitoring can conduct real-time substation and surveillance, to deal with various contingencies in a timely manner. At present, the video monitoring system in traditional substations is limited by technology developments. And it needs staff on duty for long-term monitoring, leading to the limited effect of the online monitoring system. It seriously hampered the construction of modern intelligent electric net. Therefore, it is a good idea to use computer vision in intelligent electric net monitoring system to make electrical equipment fault analysis automatically a truth. It can greatly enhance the safe operation of the substation emergency support capability.Based on the collect of insulator image data by the unmanned substation, image processing and pattern recognition methods are used to intelligently recognize the insulator abnormal circumstances in unmanned substation. The main research work of this paper are as follows:Firstly, this paper designs the methods of insulators fault diagnosis based on infrared image. According to analyzing the process of testing overheat defect of electrical equipment, which is based on the principle of infrared temperature-measuring. It analyzes temperature difference of insulator images with the use of BP network to computer the standard of temperature. And it designs a software which tests overheat defect of the insulator based on the relative temperature difference judgement.Secondly, it studies insulator image fault recognition based on machine vision. On the image denoising aspect,comparing the effect of the traditional spatial filter and pulse coupled neural network model by experiments. It shows that the method of pulse coupled neural network model is better than other methods which are on peak signal to noise ratio. And the quality of the image will be improved obviously.It used the sacle invariant features to transform operator for the matching the insulator history images. It identifies the insulator key parts by constructing scale space. Then, it accomplishes image segmentation and feature extraction based on the characteristics of the recognition sites. It can extract the feature image data by using Hu invariant which respect to transform,scaling,and rotation.Finally, accomplishes the fault insulator classification and recognition. The thesis identifies the two fault types which are the surface defect and the scar. Firstly, it discussed the identification of BP network and RBF network for the surface defect fault. The results of experiments show that the recognition of RBF network is on high rate and strong approximation ability; then, it designs the method to the fault of insulator scar that based on the detection of HSV color space pixel static algorithm. It accomplishes the insulator scar fault detection automated by non-contact.Experimental analysis shows that the insulator fault detection algorithm is proposed in this paper is accurate and provides strategic decision for real-time, automation and intelligent of power grid.