GRNN appliedtostudytheLightning Shielding
FuZhiyang · 2006
Lightning shielding failure isthemainfactor to threat thesafety andreliability ofEHV transmission lines. In thispaper,thesmall-scale modelexperiments oflightning shielding performance ofEHV transmission lines weremade, taking thevarious influencing factors intoaccount, including protection anglea ,thegradient ofthehillside 0,theheight ofthegullyH, theconductor potential U, non-vertical leaderangle4?andtheposition oflightning strokepoint (x,y) ,showing thelightning stroke probability Pahasthe hiddennon-linear function relation withthefactors above. Duetothefact thatthegeneralized regression neural network (GRNN)hasseveral advantages overtheback-propagation network (BP)suchasveryshort training timeandguaranteed performance evenwithsparse data, itisapplied tosimulate thecomplexnon-linear function relationship betweenthe lightning stroke probability Paandthesevarious influencing factors. A lotofsimulation wascarried outinMatlabandthe feasibility andaccuracy oftheproposed GRNN areverified. Index TermsEHVIUHV transmission line,lightning shielding failure, GRNN,modeltest