快速BP网络在负荷动态建模中的应用

APPLICATION OF FAST BACK PROPAGATION NEURAL NETWORKS IN POWER SYSTEM DYNAMIC LOAD MODELING

指出了负荷动态建模过程中存在的问题，提出用人工神经网络（ Ａ Ｎ Ｎ）来建立负荷的动态模型。从修改 Ａ Ｎ Ｎ 误差准则函数的角度出发，在误差准则函数中引入网络输出的能量约束，提出了一种快速误差反向传播（ Ｆ Ｂ Ｐ）算法，并利用动模和现场试验数据验证了 Ｆ Ｂ Ｐ网络用于负荷动态建模的有效性和优越性。

The im portance of dynam ic m odels of pow er system has long been recognized. And the feedforw ard perceptrontype artificialneuralnetw ork is used m ost w idely in the training m odels. This paper points out som e existing problem s in theprocess ofbuilding the dynam ic load m odeland presents a new w ay to establish pow er system dynam ic m odels w ith artificialneural netw orks ⒉ A N N⒕. The dynam ic load m odel using a fast back propagation ⒉ F B P⒕neural netw ork is developed. It isbased on the correction of its cost function by introducing an energy constraint. The sim ulation results on physicallaboratoryand field tests dem onstrate its validity and superiority in dynam ic load m odeling.