基于神经网络的电力系统负荷特性辨识_任海燕.pdfVIP

5电气开关6 ( 20 10. N o. 4) 31 : 1004 - 289X ( 20 10) 04 - 0031- 04 基于神经网络的电力系统负荷特性辨识 任海燕, 吴奋读 ( 广西大学电气工程学院, 广西 南宁 53004) : 负荷建模在电力系统分析中起着十分重要的作用参 辨识是负荷建模的关键, 好的辨识方法能够在最 短的时间内找出最优的辨识结果, 提高建模的效率首先介绍了静态负荷模型和动态负荷模型, 其次介绍了神经 网络, 最后基于神经网络对电力系统负荷特性辨识应用线性BP ( LBP )网络的参 辨识方法, 分别对静态负荷 模型(幂函 模型多项式模型) 和动态负荷模型(差分方程模型)的参 进行辨识通过现场的实测 据辨识了 模型的参 , 并验证了模型的有效性 : 电力系统; 参 辨识; 负荷模型; 神经网络 : TM 71 : B Iden tif ication of Load Characteristic of the E lectric Pow er Sy stem B ased on a N eural N etwork REN H a i-yan, WU F en-du ( College of E lectr ca l Eng neer ng, Gu angx Un vers ty, N ann ng 530004, Ch na) A bstract: The load m odel ng plays a very m po rtant ro le n the pow er system analys s. P aram eter den t f cat on s the key to load m odel ng, a good w ay for dent fy ng param eter can obta n a best resu lt n the shortest tm e, wh ch can greatly enhance the eff c ency o f the m ode l ng. T h s paper ntroduces the stat c load m odel and dynam c load m odel, then ntro- duces neura l newt ork, at last based on neural newt ork w e dent f ed pow er system load character st c. T he param eters o f stat c( exponent funct on, po lynom al) and dynam c ( d fferent a l equat ons) m odels can be dent f ed by the m ethod w h ch s used n the L near Back-P ropagat on newt ork. The m ode l param eters can be dent f ed by the data of the f e ld, and ver fy the e ffect veness o f the m ode.l K ey wo rd s: pow er system; param eter dent f cat on; load m odels; neural netw ork 1 2 , , , , , , : [ 1]