The Study On Monitoring Method Of Substation Equipment Based On The Technology Of Dual Video Information Fusion

Infrared thermography as a fast, non-contact monitoring tool has been widely used in electric power industry, The temperature measurement of early hand-hold infrared thermal camera need to be carried out by manual inspection.Due to the increasing of equipment malfunctions caused by internal thermal defects,the demand for fully automatic monitoring system was becoming even more prominent. In our country, the first generation robot for substation equipment inspection was created in 2002, and it has been rapidly universalized in recent years. This type of systemcarry both infrared and visible sensor, while related monitoring systems have also developed at the same time. Based on these developments, this thesis proposes a method of dual video information fusion method and designsa transformer equipment condition monitoring system with this method.The thesis is going to extract linear edge features to implement dual video images registration. First the principle of duality is used to deduce the theory for calculating the transformation matrix,based on which straight edges are found in the image.Then set up three parameters of each line and find the matching pairs between the two source images, after at least four pairs of lines was achievedthe transformation matrixcan be calculated. Compared with the typical SURF and BRISKfeature extraction algorithms, the proposed algorithm is better for the extraction effects of infrared image, which could be more simple and effective when it is applied to substation equipment.The dual video images fusion algorithm is based on sub-band decomposed multi-scale retinex(SD-MSR), whichhas been used for image enhancement methods widely.This paper has improved it by proposing complementary intensitytransfer functions.By employing two different intensitytransfer functions, we can enhance details in bothshadows and highlights.The fusion method processes image enhancement and image fusion steps at the same time. In the experiment, we compared this method withthe method based on gradient fusion and based on wavelet fusion. The result of subjective evaluation indicates that the method has better contrast and more details, and it also shows that both of the average gradientand the spatial frequency are better than the other two methods, which means that the method is better than the othertwo fusion methods in terms of image sharpness and details.Asubstation equipment state monitoring system is designed, and the hardware of this system uses a pillar-type dual video monitoring device whichdoes not needto recharge and can run all the time. Whilethe software of this system applies the algorithm to achieveregistration and information fusion of dual video images, so that the system can get more informationin complex weather environment and avoid errors alarm or not alarmdue to image quality declined. Moreoverthis system can be also used for monitoring such burst events asextraneous body breaking into, burning,explosion and so on.