Tattalin Tadakawa na Tashin Kula da Tashin Karamin Tsakiyar Gwamnatuwar Yawan Dajiya a Najeriya Da Tasiri a IEE-Business Ta Hanyar ƙarfin Hotuna Na Zafiya

Mafiyanin jirgin kofin zama (a nan da yake mafiyanin jirgin kofin) suna da amfani a tsarin jirgin zama saboda muhimmanci masu kyau kamar inganci, kadan kadan, ba a yi gaba-gaban hoton zuwa kai, tattaunawa, kadan sauti, kadan hankali na kofin, lokaci mai kawo kofin, da kuma kadan bayanin al'amuran. Idan hoton zuwa kai yana ci gaba da kusa, a lokutan kasa kamar kafin samar da kisa, kisan kuli, fuskantar kisa, ko kuma rufe, za su iya shafi shiga cikin kofin mafiyanin jirgin kofin. Wannan ya zama da ya iya haifar da kofin mafiyanin jirgin kofin, kuma ya iya haifar da tattalin kasa da kusa da al'amuran al'adu na gwamnati.

A nan da yake, ZW32 - 12 mafiyanin jirgin kofin na musamman da ke kan waje ta duniya ta da kofin kula-kula (a nan da yake HV ZW32 - 12 mafiyanin jirgin kofin) ya zama misal, wanda ya fara da abubuwan nuna da ita a wurareen hawa na musamman. An fara da imajin UV don nuna tattalin kasa na kofin mafiyanin jirgin kofin, kuma a fara da kaddamar da kasan tattalin kasa. Ba a yi karfin bayanai na imajina UV, an fara da kaddamar da muhimmancin waɗannan imajina. Sannan, an fara da kaddamar da kasan tattalin kasa ta hanyar sadarwa masu lissafi support vector machine, wanda ya ba da ita a nufin imajina UV. Wannan ya zama yanayin bayanai mai sauƙi mai kasa na tattalin kasa na mafiyanin jirgin kofin.

ZW32 - 12 mafiyanin jirgin kofin shine mafiyanin jirgin kofin na tsohuwar adadin 50Hz, 12kV AC na musamman da ke kan waje ta duniya. Ana amfani da shi don kawo kofin da kuma kofin adadin kasa, kofin adadin kasa mai yawa, da kuma kofin adadin kasa mai yawa. Tana da taswira a Fig. 1.

Don kawo kofin imajin UV na tattalin kasa na kofin mafiyanin jirgin kofin da kuma kaddamar da kasan tattalin kasa (PD), an fara da sistem na bayanai na tattalin kasa na kofin, kamar yadda aka nuna a Fig. 2. A Fig. 2, T shine tashin kudeta, B shine tashin kudeta mai kasa, R₁ shine tashin kudeta mai kasa, da C₂ shine tashin kudeta mai kasa, wanda ana amfani da shi don kawo kofin PD.

Tashin kudeta da ake amfani da shi a cikin sistem shine YDWT - 10kVA/100kV model, kamar yadda aka nuna a Fig. 3 - a. Ana amfani da shi don kawo kofin sursurin kudeta da ke buƙata na kofin.

An fara da OFIL Superb UV imager don kawo kofin imajin UV na tattalin kasa na kofin, kamar yadda aka nuna a Fig. 3 - b. Samfurin bayanai shine kofin mafiyanin jirgin kofin na ZW32 - 12, wanda ya yi aiki har shekaru uku, kamar yadda aka nuna a Fig. 3 - c. An fara da samfurin a cikin tashin kudeta na musamman, inda za a iya kawo kofin adadin kasa mai kyau.

A cikin wannan sistem, an fara da yanayin pulse current don kaddamar da kasan tattalin kasa (PD). Kontrola ta kawo kofin tashin kudeta da tashin kudeta don kawo kofin sursurin kudeta. Sannan, an fara da signalin PD zuwa JFD - 3 PD detector ta hanyar tashin kudeta mai kasa da kuma tashin kudeta mai kasa.

Idan an yi kasa da kasa, za a iya kawo kofin adadin kasa mai kyau a cikin tashin kudeta. Kofin mafiyanin jirgin kofin suna da kasa da sursurin kudeta har biyu don kawo kofin adadin kasa mai kyau. Sannan, an fara da sursurin kudeta 12kV zuwa kofin mafiyanin jirgin kofin da kasa biyar da minuɗu. A cikin wannan lokaci, an fara da imajin UV, kuma an fara da kaddamar da kasan tattalin kasa. Daga imajin UV shine 5 mita, da daraja 0°, da kuma gain 110%. An fara da bayanai a cikin adadin kasa duka, kamar yadda aka nuna a 70% zuwa 90%, da kasa da 5%.

 Tsarin imajina UV

Imajin UV an fara da videon, kuma ya kamata a yi karfin frame don samun frames na imajin UV don bayanai. Har frame na imajin shine RGB true-color image [3]. Tattalin kasa na kofin an fara da ita a cikin imajin UV a matsayin manzara mai kyau. Idan tattalin kasa na kofin yana da takaitaccen, yana da manzara mai kyau. Saboda haka, ya kamata a yi karfin bayanai da kuma karfin bayanai na imajin don kawo kofin background na imajin da kuma kawo kofin manzara mai kyau.

Saboda red component (R), green component (G), da blue component (B) a cikin RGB color space na nuna ratio na color da kuma ba na iya nuna brightness na imajin, ana fara da kawo kofin har frame na imajin a HSL color space. HSL na nuna Hue, Saturation, da Luminance respectively. HSL components na imajin frame an fara da ita a Fig. 4. Daga Fig. 4, za a iya cewa H ko S component ba na iya kawo kofin manzara mai kyau daga background, amma L component na iya kawo kofin [4].

Daga Fig. 4 - c, L component na manzara mai kyau yana da karamin background. Saboda haka, threshold segmentation shine yanayin bayanai mai sauƙi don kawo kofin manzara mai kyau. Abubuwan da ya kamata shine zama threshold na L-component. A nan, ana fara da Otsu's thresholding method don kawo kofin L-component threshold [5]. Ba a yi Matlab coding don Otsu's method, an fara da L-component threshold da ya danganta 216, da kuma result na segmentation an fara da ita a Fig. 5 - c. Da ya gane, background an kawo kofin, da kuma manzara mai kyau na imajin UV an fara da ita.

Daga Fig. 5 - c, akwai noise points na biyu. Don kawo kofin abubuwan, ana fara da mathematical morphology operations ta hanyar structural element na circle da radius 4 pixels don kawo kofin noise points [6]. Ba a yi mathematical morphology processing, an fara da result a Fig. 5 - d. Duk noise points an kawo kofin, da kuma manzara mai kyau na imajin UV an fara da ita. Ana ce area na manzara mai kyau na imajin UV "facula area" na imajin UV.

Ba a yi kaddamar da facula area na frames na imajin UV, an fara da curve na facula area. Curve na facula area a 85% humidity an fara da ita a Fig. 6. Daga Fig. 6, facula area na yake da karamin small range, da kuma manzara mai kyau na biyu. Saboda haka, an fara da three parameters don kawo kofin intensity na tattalin kasa: mean facula area, intermittent facula area, da repetition times of intermittent facula respectively [7]. Ana fara da 100 consecutive frames ba a yi aiki don kawo kofin. Mean facula area shine average na areas na 100 frames' faculae. Intermittent facula area shine average na areas na faculae da suka da area mai kyau da mean facula area, da kuma repetition times of intermittent facula shine number of faculae da suka da area mai kyau da mean facula area. Daga Fig. 6, mean facula area shine 665 pixels. Intermittent facula area shine 902 pixels. Repetition times of intermittent facula shine 32.

Idan an fara da three characteristic parameters da kuma kaddamar da kasan tattalin kasa (PD) synchronously, an fara da kaddamar da kasan tattalin kasa (PD) ta hanyar three UV image parameters ta hanyar least-square support vector machine method.

An fara da 90 samples na imajin UV. Ta hanyar har frame na samples, an fara da three UV image parameters, da kuma kaddamar da kasan tattalin kasa (PD) an fara da ita zuwa JFD3 PD detector. Input arguments na vector machine an fara da mean facula area, intermittent facula area, repetition times of intermittent facula, da relative humidity. Output argument shine kasan tattalin kasa (PD). Radial Basis Function (RBF) an fara da ita ta hanyar kernel function. Ba a yi normalization, 80 samples an fara da ita don training. Kernel parameters da punishment parameters na vector machine an fara da ita ta hanyar default values. Result na training an fara da ita a Fig. 7.

Daga Fig. 7, duk training samples, error na measured PD quantity yana da karamin small. Amma, wasu samples, error yana da karamin 20%. Mean Square Error (MSE) an fara da ita ta hanyar:

Don kawo kofin Mean Square Error (MSE) na regression result da kuma zama accuracy na vector machine, genetic algorithm (GA) an fara da ita don optimize kernel parameters da punishment parameters. [8 - 9]

Termination generation an fara da ita ta 100, da population size ta 20. Optimization process an fara da ita a Fig. 8. Daga Fig. 8, ba a yi 30 generations of evolution, MSE an fara da ita ta 0.07 zuwa 0.01, wanda ya nuna cewa genetic algorithm an samu optimal point. [10] Optimized kernel and punishment parameters an fara da ita 0.2861 da 82.65 respectively.

Ba a yi optimization of the parameters using the genetic algorithm (GA), an fara da same 80 samples retrained, da kuma regression result an fara da ita a Fig. 9. Daga Fig. 9, duk samples an fara da error na biyu da measured partial discharge (PD) quantity. Mean Square Error (MSE) an fara da ita 10, wanda ya nuna cewa ita ya da karamin small da value of 80 before the parameter optimization. Saboda haka, an fara da cewa optimizing the GA parameters can effectively reduce the MSE of the regression result and enhance the accuracy of the vector machine.

Final 10 samples an fara da ita don conduct a test on the model. Regression results an fara da ita a Table 1. It can be clearly observed that the error between the regression results and the actual partial discharge (PD) quantity is less than 6.1%. This finding indicates that the trained model demonstrates excellent generalization ability.

UV imaging technology an fara da ita don detect the surface discharge of outdoor vacuum breaker post insulators. The relationship between the facula area in UV images and the partial discharge quantity is explored through the least-square support vector machine method, offering a novel approach for insulation fault diagnosis of outdoor vacuum circuit breakers based on ultraviolet imaging.

A. Image Processing and Parameter Definition

Ba a yi L-component threshold segmentation da mathematical morphology operations na imajin UV, an fara da spot part na imajin UV, don kawo kofin facula area. Three parameters an fara da ita don quantify the discharge intensity: mean facula area, intermittent facula area, da repetition times of intermittent faculae.

B. Data Acquisition and Analysis

Idan an fara da imajin UV da kuma kaddamar da kasan tattalin kasa (PD) synchronously, an fara da relative humidity da three UV image feature parameters as input variables. Through regression analysis via the least-square support vector machine, along with kernel parameter optimization using a genetic algorithm (GA), the PD quantity can be accurately determined.

C. Diagnostic Accuracy and Significance

By conducting regression analysis to establish the relationship between the insulator surface discharge quantity and its UV image facula area, it is found that the PD quantity diagnosed solely from UV images has an error of less than 6% compared to the measured PD quantity. This level of accuracy meets the requirements of practical applications and provides a new non-invasive method for diagnosing external insulation faults in outdoor vacuum circuit breakers based on ultraviolet imaging.

This research was funded by the National Natural Science Foundation of China and the State Key Laboratory of Electrical Insulation and Power Equipment. The authors would like to express their sincere gratitude to all those who provided support for this project.