基于MC-ANN的中性点直流监测数据有效性评估

孙冰; 何常根; 孙谊媊; 王开科; 赵普志; 吴伟丽

孙冰,何常根,孙谊媊,王开科,赵普志,吴伟丽.基于MC-ANN的中性点直流监测数据有效性评估[J].电测与仪表,2020,57(15):88-94.
SUN Bing,HE Changgen,SUN Yiqian,WANG Kaike,ZHAO Puzhi,Wu Weili.Effectiveness evaluation of neutral DC monitoring data based on MC-ANN[J].Electrical Measurement & Instrumentation,2020,57(15):88-94.

基于MC-ANN的中性点直流监测数据有效性评估

Effectiveness evaluation of neutral DC monitoring data based on MC-ANN

DOI：10.19753/j.issn1001-1390.2020.15.014

中文关键词: 性点直流监测数据影响因素神经网络蒙特卡洛模拟

英文关键词:Neutral point DC Monitoring data Influence factors Artificial neural network Monte Carlo simulation

基金项目:国网新疆电力公司电力科学研究院直流偏磁资助项目（5230DK17000K）；新疆维吾尔自治区科技厅面上基金（2017D01C417）

作者中文名	作者英文名	单位
孙冰	SUN Bing	国网新疆电力有限公司电力科学研究院
何常根	HE Changgen	国网新疆电力有限公司
孙谊媊	SUN Yiqian	国网新疆电力有限公司
王开科	WANG Kaike	国网新疆电力有限公司电力科学研究院
赵普志	ZHAO Puzhi	国网新疆电力有限公司
吴伟丽	Wu Weili	西安科技大学

摘要点击次数: 1665

中文摘要:

针对特高压直流输电工程试运行期间得到的变电站中性点直流监测数据,提出了数据有效性评估方法：首先,根据变电站中性点直流分布机理,分析了影响因素与变电站偏磁电流分布之间的关联特征,构建了基于神经网络(Artificial Neural Network,ANN)的中性点直流预测模型,并与实测数据进行对比,验证预测模型的有效性；接着,利用蒙特卡洛模拟(Monte Carlo,MC)法对各影响因素进行抽样,模拟各种影响因素不确定组合,作为输入送入已经训练好的神经网络预测模型,构建MC-ANN联合模型,获取影响因素抽样下的大量样本数据,利用数据挖掘技术提取影响因素与中性点直流之间的关系变化规律；最后,针对±800kV上海庙-山东临沂特高压输电工程测试数据进行有效性评估,结果表明采用本文方法能够快速查找到测试期间中性点发生变化的站点。本文方法可为甄别异常监测数据、快速查找异常原因以及锁定接地方式变化站点提供理论参考。

英文摘要:

Aiming at the neutral point DC monitoring data of the substation obtained during the trial operation of UHV DC transmission project, the data validity evaluation method is proposed. Firstly, according to the neutral distribution DC distribution mechanism of the substation, the influencing factors and the substation bias current distribution are analyzed. Based on the correlation feature, a neutral point DC prediction model based on Artificial Neural Network (ANN) is constructed and compared with the measured data to verify the validity of the prediction model. Then, Monte Carlo (MC)simulation is utilized. The method is to sample the influencing factors, simulate the uncertain combination of various influencing factors, input the trained neural network prediction model as input, construct the MC-ANN joint model, obtain a large number of sample data under the influencing factors, and use the data. The relationship between the influencing factors of mining technology extraction and the neutral point DC is studied. Finally, the validity of test data of ±800kV Shanghai Temple-Shandong Linyi UHV transmission project is evaluated. The results demonstrate that the method can quickly find the test period. Site where the point of change has changed. This method can provide a theoretical reference for screening abnormal monitoring data, quickly finding the cause of the anomaly, and locking the grounding change site.

查看全文查看/发表评论下载PDF阅读器

关闭