| The railway is an important infrastructure in China.It has become a widely used transportation mode with its own economy and popularity,and it is in the backbone of China’s transportation system.The reliable and stable operation of the electrified traction power supply system,which is the core of the high speed railway,is an important prerequisite for the safe production of railway transportation.China’s Xinjiang Province is a typical arid and semi-arid climate.In this region,Gobi and dry Saline Lake are widely distributed.There are many windy areas,which often lead to typical salty and dusty weather in this area.In short time,the insulator surface of the contact network can attach to the thick filth,then causes the frequent pollution flashover in the wet environment,which seriously affects the safe production of railway transportation.Therefore,in the meteorological environment with high concentration of atmospheric particles and high salt content,it is necessary to study the prediction method of pollution flashover suitable for this area based on the available meteorological data.In this thesis,the general dynamic sedimentation model of the pollution particles on the insulating surface is firstly established.Based on the meteorological characteristics of saline-alkali dust that are unique to the Xinjiang region,the dynamic accumulation is divided into continuous accumulation during no rainfall and dynamic erosion during rainfall.Secondly,a settling model of pollution particles based on fluid mechanics was established.Combined with the particle size distribution of each site downstream of the pollution source area and the different gray-salt ratios at each site,the real-time salt density and ash density values were calculated using the accumulative fouling formula during continuous contamination.Then,the characteristic parameters of rainfall were analyzed.The parameters such as monthly average rainfall intensity,monthly average wind speed,number of rainfalls,salt density in this month,and gray density in this month were used as input features to establish a rainfall erosion model based on genetic BP neural network.This model accurately map the relationship between rainfall parameters and pollution parameters.Finally,the early warning voltage is set according to the information of the numerical weather forecast system,the influence of the salt density and the ash density on the pollution flashover voltage,and the uniform creepage distance set by the pollution area distribution level.The response mechanism is determined by whether the withstand voltage at any time obtained by ESDD and NSDD is up to the early-warning voltage.To clean all the insulators in the area within the specified time will be changed to clean insulators which meet the pollution flashover condition before the pollution flashover weather is coming,so as to avoid the waste ofmanpower,material and financial resources and prevent the occurrence of pollution flashover at the greatest extent. |
