Research And Evaluation On The State Characteristics Of The Whole Process Of Partial Discharge Development Of Oil Paper Insulation

Partial discharge is the main form of deterioration of transformer oil paper insulation,and the long-term existence of partial discharge may further develop the insulation deterioration,which can lead to insulation breakdown or flashover.Therefore,an accurate diagnosis by partial discharge detection of transformer insulation condition,and the dangerous defects timely warning,are important goals of the power department,which carries out the state operation maintenance,are also the key technical problems in the field of the current state of partial discharge detection.Compared with Dissolved Gas Analysis(DGA),the detection method of partial discharge can detect the defects of rapid development,which has high sensitivity and good real-time performance.Therefore,it can be used to detect the dangerous discharge condition of oil-paper insulation by the detecting information of partial discharge.This paper uses the method of constant voltage and the real-time HF/UHF detection system to collect and analyze the signals,which are from starting to breakdown of the whole process,of the oil-immersed board discharge and the oil wedge discharge in transformer oil-paper insulation.The trend and regularity of the amplitude,frequency,entropy,synchronous joint quantity of the signals and PRPD/PRPS spectra of the oil-immersed board discharge and the oil wedge discharge are studied,the whole process is divided into the stable development stage and the accelerated deterioration stage,and the state characteristics of oil-immersed board and oil wedge discharge in two phases are analyzed and summarized.The influence of voltage level on the discharge development process is studied by changing the voltage level,the change rule of statistic parameters of the whole process is also summarized under different voltages.Based on the development trends of statistic parameters,which reflect the various conditions of discharge stage,it is constructed of 27 features,using Principal Component Analysis(PCA)and Fuzzy C-Means algorithm(FCM)for cluster analysis of the discharge stages,and it further validates the rationality of the two stages.Then the discharge mechanism,energy characteristics of different signals and discharge characteristics of different types are explained from the angle of mechanism.Finally,considering the on-site partial discharge detection,the characteristics are explored,which can effectively distinguish between the stable development stage and accelerated deterioration stage.An effective method,which can determine the development stage and early warning of oil-paper insulation board,is presented.The results show whether there are the trends and regularity of the amplitude,frequency,entropy of the HF/UHF signals,or the machine recognition results of PCA and FCM algorithm,the whole process can be divided into the stable development stage and the accelerated deterioration stage.Under the constant voltage,the transition from the stable development stage to the accelerated deterioration stage has obvious transition process.Under different voltages,the overall trend of discharge development is basically the same,and there are some differences.Finally the characteristics of two phase-average deviation value ?N??V??(?t)are explored,and the stage recognition used BP neural network has good effect,which can be used as the characteristic of partial discharge condition monitoring and warning.The research results of this paper provide a test basis for the state monitoring of the transformer oil paper insulation,and have the engineering guiding significance for the warning of dangerous faults.