Research On Power Grid Contaminated Flashover Prediction System

It is a disaster that threatens the safty of the power grid when contaminated insulators flashover occurs. Along with the continuous development and scale increasing in grid, the damage caused by contaminated insulators flashover is more and more serious. Monitor the status of contaminated flashover, prevent and control the power grid flashover disaster is an important job of electric power operating and management department. It is one of the hotspots and difficulties on which the electric power administration and relative research institutes focus that the accurately forecast of the flashover occurence.A power grid contaminated flashover forecasting model is proposed in the thesis on the foundation of practically and accurately flashover forecasting and warning theory study. The background of the thesis is based on the project "Research and Development of Comprehensively Disaster Prevention and Mitigation of Power Grid" from State Grid Corporation. The focal point of the thesis is building up a method of contaminated flashover to which meet the practically engineering requirments of electric power industry.The proposed model with multiple structures can forecast the equivalent salt deposit density and cirtical flashover voltage, then estimates the status of flashover level and give the classified pre-warning.The main contents of the thesis can be concluded as follow:Analysis the characteristic of the equivalent salt deposit density measuring series that is susceptible to meteorological and geographic factors, study the equivalent salt deposit density nonlinear time series and its forecasting method. Study the features of the quivalent salt deposit density time series and its characteristic of nonlinear and chaos, the method of phase space reconstruction and the way to determine its parameters of delay time, embedded dimension and max Lyapunove exponent is determined. Based on phase space reconstruction of chaotic time series, study the globe equivalent salt deposit density nonlinear time series forecasting model is presented. For the reason that equivalent deposit density time series is a small sample size series, a support vector machine algrithom to fit the nonlinear mapping in reconstructed equivalent salt deposit density system phase space is proposed in the thesis, and the results of equivalent salt deposit density support vector machine nonlinear time series forecasting model is acceptable.Based on artificial pollution voltage withstand test and analysising of the data , the test data is used to build the forecasting model that fit the multidimensional nonlinear relationship among the flashover critical state voltage, equivalent salt deposit density and meteorologic factors. On the base of the multi-dimensional nonlinear relationship between crictial pollution flashover voltage and equivalent salt deposit density and environment meteorology factors, the critical flashover voltage ANNs forecast model is built. Based on the nonlinear relationship between crictial pollution flashover voltage and operating voltage, build up the flashover critical state forecast model that can forcast the pollution flashover state using the real time weather forcast data.On the basis of the forecast and pre-warning mechanism of power grid contaminated flashover, the power grid contaminated flashover predition system is proposed and it application software based on Java technology and Oracle database technology is developed. The proposed mechanism can classify the flashover risk and the level of pre-warning, and the forecast of contaminated flashover based on it has practical value to guide the engineering practice of contaminated flashover prevention and mitigation.The developed contaminated flshover predition system had been applied in practical engineering of power grid operation management after passing the inspection and acceptance. Its performance and running index meet to the demand of the designing and satisfied to the engineering requirements of the power grid operating and management. The built power grid contaminated flashover predition system was brought into service in April, 2008, and operating practice indicates that it works satisfactory and the result of forecasting and pre-warning is accteptable in practical engineering application for over one year's service.