Study On Methods Of Recognition Feature Extraction And Fractal Compression For Partial Discharge Gray Intensity Images

Partial discharge (PD) inside insulation is considered as one major cause of insulation degradation in electrical equipment and attached importance to the safety and reliability of running electrical equipment. Auto-recognition to discharge types in on-line PD monitoring system could be used to find out internal partial defects and the relevant discharge development degree in time, and then prevents equipment from the coming faults. According to the requirements to PD pattern auto-recognition, this paper studies systematically the basic theories and realizable methods for auto-recognition of PD gray intensity image: (1) Aiming at extraction of PD fractal features, we study the fundamental fractal theory and specific methods for fractal dimension evaluation, and then bring forward the modified differential box-counting (MDBC) method for the first time. Meanwhile, fractal dimensions of fractal Brown motion images and texture images are computed by MDBC method and results testify that the proposed method brings higher precision in fractal dimension evaluation and reflect better the gray intensity distribution of those images. With the MDBC method, we established the foundation for research on extraction of PD fractal features. (2) In the requirement of on-line PD monitoring for transformer, several discharge models are designed and the relevant experiment methods are projected. With discharge model tests, a lot of discharge sample data is acquired. On the base of systematical research on recognition for PD gray intensity image, this paper puts forward two kinds of fractal features, the 2nd generalized dimensions of original PD images and fractal dimensions of high gray intensity PD images, and then the relevant extraction methods. Meantime, moments and correlative statistical features are studied for recognition of PD gray intensity images. Moreover, it's the first time to put forward and study the method to use recognition feature set consisting of above three kinds of features for auto-recognition of PD pattern. (3) In this paper, the method of quadtree partitioning fractal image compression (FIC) is studied and used for PD gray intensity image compression. The simulation test results show that determinate compression ratio is achieved by the algorithm proposed in this paper. Furthermore, good compression effectivity is presented in application to compression of PD gray intensity images. According to the research on difference degree between computational values of fractal features extracted from decoded PD images and that from original images, it is shown elementarily that the proposed method is effective for application in PD pattern auto-recognition system. (4) Furthermore, PD pattern auto-recognition project is brought forward on the base of the above recognition features and fractal compression of PD gray intensity images and then the classifier is designed with back-propagation neural network (BPNN). The comparatively high recognition correctness probability is achieved in classification to original PD images. According to furthermore researches on recognition of decoded PD images, it is testified and meanwhile shown that the designed project for PD pattern auto-recognition is effective and can meet requirements for PD pattern auto-recognition in field and PD data telecommunication and remote auto-recognition. The above research results show that the proposed PD recognition features set and FIC method, together with BPNN classifier, can be effectively used for PD pattern recognition and good recognition results are achieved.