Study On The Forecasting Characteristics And Method Of Flashover Of Ice-covered Insulator

As an important part of the transmission lines, Insulators are widely used in powersystem. Ice disaster occurred frequently due to iced insulators of transmission lines,which posed a serious threat to the security and stability of power system operation.Nowadays numerous studies have been conducted on the icing flashover mechanisms,however, there are rarely researches on ice flashover warning and the well-recognizedcharacteristic parameters to assess the iced insulators’ operating state haven’t beenproposed. So in this paper，based on a large number of artificial icing tests on110kVcomposite insulators, leakage current and ultraviolet photon number under operatingvoltage in different icing conditions in the process of melting ice are collected toextract the characteristic parameters of leakage current that accurately indicate theoperating state of iced insulators. The study investigates the relationship between thesalt density, ice amount and characteristic parameters and then establishes twoevaluation models to predict the ice flashover voltage. Finally, a method of icedinsulators hierarchical flashover early warning is proposed.The main achievements are as follows:①In this paper,110kV composite insulators icing tests in total30icing conditionsin consideration of five kinds of icing degree, six kinds of salt density are carried out inthe multifunctional artificial climate chamber. In order to extract the characteristicparameters of the leakage current effectively reflecting the state of iced insulators andthus to obtain the relationship among them, the changing process of leakage current inthe process of melting ice is analyzed as well as the impact of salt density and iceamount on leakage current.②The ultraviolet photon number of110kV iced insulators in different degree ofice is collected and the change of the ultraviolet photons in the ice melting process isanalyzed. Then the study extracted the characteristic parameters of the ultravioletphoton that effectively reflect the degree of discharge and the average number ofultraviolet photon photons (N). The relationship between the salt density, ice amountand the average number of ultraviolet photon (N) is given. According to the flashovervoltage experiments，Ice flashover voltages under different salt density and ice amountare obtained so as to fit the relationship among them. ③The paper puts forward two kinds of iced insulators flashover voltageevaluation models, which are the generalized regression neural network (GRNN)model and minimum support vector machine (LS-SVM) model. By trainings andexperiments, the accuracy of the models is verified well. The results show thatmaximum error of two modes is9.04%and9.04%respectively. After that，a method oficed insulators hierarchical flashover early warning method is proposed.