摘要
冰冻灾害下覆冰易于造成断线倒塔等电力事故,提出一种基于改进粒子群算法优化NRBF神经网络的覆冰厚度预测模型.通过改进粒子群算法,优化最近邻聚类算法的聚类半径,确定NRBF神经网络隐含层节点个数,并运用优化后的神经网络对覆冰厚度进行预测.以2006年湖南电网220kV黔平线路的覆冰数据为例,分析验证了该模型的合理性,为输电线路防冰、除冰提供理论依据.
Transmission line and towers are easily broken down and collapsed by icing, an ice thickness prediction model based on improved particle swarm algorithm and normalized radial basis function （NRBF） neural network is proposed in this paper. The ice thickness of transmission lines is predicted by ameliorated NRBF neural network, and the parameters of NRBF neural net- work is determined through improved particle swarm algorithm to optimize the cluster radius of nearest-neighbor clustering. Taking the icing data from 220 kV Qianping lines in 2006 as an ex- ample, analysis results show that the prediction mQdel is effective and feasible, which provides a theoretical basis for anti-icing and deicing.
出处
《电力科学与技术学报》
CAS
2012年第4期76-80,共5页
Journal of Electric Power Science And Technology
关键词
冰冻灾害
覆冰厚度
最近邻聚类算法
改进粒子群算法
NRBF神经网络
freeze disaster
ice thickness
nearest-neighbour clustering
enhanced particle swarm algorithm
NRBF neural network
