Microscopic Process Simulation And Pulse Characteristic Analysis Of Typical Partial Discharge

Ultra-high frequency(UHF)detecting method for Partial discharge(PD)is widely used in online detection due to its great advantages,such as high detection sensitivity,suitable for various discharge types,non-touching measurement,and so on.Study the microscopic principle and research the transient current pulse characteristics of PD,and further study the excited electromagnetic pulse signal and its propagation characteristic are the theoretical basis of PD UHF detecting method and important precondition to promote the application of UHF method and discovery of new method in PD detection.With the basis of investigating PD fluid model and kinetic model,this dissertation takes advantage of particle-in-cell Monte Carlo collision(PICMCC)method,which can directly analyze the movement process of charged particle,to simulate the microscopic process of PD.Air-gap discharge,point discharge and surface discharge are selected as three kinds of typical PDs for modeling and carry on simulation.The micro forming process of PD is intuitively observed and the current pulse signal of three kinds of PD is obtained.This dissertation considers the quantitative change rule of charged particle on different place at different time,mathematically fitting discharge pulse waveform through a novel sectionalized method.In this way,the function is obtained to reflect the physical process,and the fitting degree reaches higher than 96%.By changing physical parameters of simulation model,the amplitude,rising time,falling time and pulse width are quantitatively analyzed,and the variation tendency of waveform characteristics with the change of physical condition is further confirmed.Moreover,owing to the experimental requirement,a portable partial discharge generator is developed,including programmable high voltage power supply module and detachable PD model module.Several experiments are carried out by using this portable PD generator and the experimental data of air gap discharge and point discharge is obtained.Generative adversarial networks(GAN)are introduced for PD pattern recognition.