摘要
分析了影响风电场发电量的主要因素,提出了一种新的基于人工神经网络的风电场发电量预测方法。针对传统神经网络预测模型的预测质量由于陷入局部最优、训练难收敛等原因而降低的情况,该方法引入了自适应变异粒子群算法(AMPSO)对神经网络的权值和阀值进行训练,并且在训练过程中通过比较不同隐层节点数所对应的输出误差来确定神经网络的最优拓扑结构。使用张北县风电场的实际测量数据进行建模预测,结果证明了本文描述的模型是一种有效可行的风电场发电量预测方法。
This paper analyzes the main influencing factors for the wind energy of wind farms,and proposes a novel model for the prediction of wind energy based on artificial neural network.Compared with traditional neural networks based forecasting methods,this model has introduced Adaptive Mutation Particle Swarm Optimization（AMPSO） algorithm to the training phase of the neural network,and by error and trial the topology of neural network is determined.Simulation and test results of this proposed network are reported,showing that the predicted values of wind energy are in agreement with the actual values.
出处
《华东电力》
北大核心
2011年第5期797-802,共6页
East China Electric Power
基金
国家863计划项目(G20080188)
浙江省科技计划项目(G20080188)~~
关键词
风力发电
发电量预测
自适应粒子群算法
人工神经网络
Wind energy
Power prediction
Adaptive Mutation Particle Swarm Optimization
artificial neural network
