摘要
安全裕度低的电网关键断面是电网运行人员需重点关注的电网薄弱环节,对其进行准确预测是保证电网安全、稳定运行的重要技术手段。以广东电网为例,收集了该地区2014和2015年的电气量和气象数据。首先,将电气量与气象数据进行标准化和集成;其次,对特征全集进行特征选择,并利用神经网络模型进行训练,得到关键断面的神经网络预测模型。相比于传统方法,所提预测模型在电气量因素的基础上,引入了非电气量因素(气象因素),用以挖掘2种因素对电网安全运行中关键断面的影响。广东电网的算例测试表明,该模型预测准确性好、速度快,适应于复杂多变的实际电网。
Critical interfaces with low safety margin are weakness of power grid need to be focused on. Its accurate prediction is important basis of reasonable scheduling. Taking Guangdong power grid for example, data of electric and non-electric parameters between 2014 and 2015 were collected. Firstly, the parameter data were standardized and integrated. Secondly, whole characteristics were selected and samples were trained with neural network. Then a neural network prediction model for critical interfaces was obtained. Compared to traditional method, this method introduces non-electric parameters. Test on Guangdong power grid shows that this model is accurate, fast and suitable to actual complex and changeable power grid.
出处
《电网技术》
EI
CSCD
北大核心
2016年第11期3399-3405,共7页
Power System Technology
基金
上海市科委科技创新项目(14DZ1201602)
南方电网科技项目(GDKJ00000058)~~
关键词
气象因素
神经网络
电网安全分析
关键断面
大数据
meteorological factors
neural network
grid security analysis
critical interface
big data