基于深度學(xué)習(xí)與諧波譜相關(guān)分析的臺(tái)區(qū)識(shí)別
徐曉東1,呂干云1,魯濤1,吳啟宇2
(1 南京工程學(xué)院 電力工程學(xué)院,江蘇 南京 211167;2 國(guó)網(wǎng)江蘇省電力有限公司南京市溧水區(qū)供電分公司,江蘇 南京 211200)
摘 要:為了提高用戶(hù)臺(tái)區(qū)識(shí)別的效率和精度,提出了一種基于深度學(xué)習(xí)與諧波譜相關(guān)分析的臺(tái)區(qū)識(shí)別方法。采集配變出口電壓進(jìn)行諧波頻譜分析,并通過(guò)深度置信網(wǎng)絡(luò)(DBN)的特征提取模型自適應(yīng)提取配變電壓特征諧波譜。提取用戶(hù)端智能電表的電壓特征諧波譜,利用譜相關(guān)分析法計(jì)算智能電表與配變間電壓特征諧波譜的皮爾遜相關(guān)系數(shù),進(jìn)而通過(guò)譜相關(guān)程度對(duì)比判斷用戶(hù)所屬臺(tái)區(qū)和相別。選取南京市某低壓配電網(wǎng)進(jìn)行現(xiàn)場(chǎng)測(cè)試,實(shí)測(cè)結(jié)果表明,所提方法提高了用戶(hù)臺(tái)區(qū)和相別識(shí)別效率,為電網(wǎng)公司對(duì)臺(tái)區(qū)精細(xì)化管理提供新技術(shù)。
關(guān)鍵詞:深度學(xué)習(xí);特征諧波譜;諧波譜相關(guān)分析;臺(tái)區(qū)識(shí)別
中圖分類(lèi)號(hào):TM715 文獻(xiàn)標(biāo)識(shí)碼:A 文章編號(hào):1007-3175(2020)08-0007-05
Transformer Area Recognition Based on Harmonic Spectrum Correlation Analysis and Deep Learning
XU Xiao-dong1, LYU Gan-yun1, LU Tao1, WU Qi-yu2
(1 School of Electric Power Engineering, Nanjing Institute of Technology, Nanjing 211167 , China;
2 Nanjing Lishui District Power Supply Branch of State Grid Jiangsu Electric Power Co., Ltd, Nanjing 211200, China)
Abstract: In order to improve the efficiency and accuracy of user transformer area recognition, a transformer area recognition method based on deep learning and harmonic spectrum correlation analysis is proposed. Firstly, collected the output voltage of the distribution transformer to analyze the harmonic spectrum, and the characteristic harmonic spectrum of distribution transformer voltage is extracted adaptively by the feature extraction model of depth confidence network (DBN). Then, the voltage characteristic harmonic spectrum of the smart meter is extracted, and the Pearson correlation coefficient of the voltage characteristic harmonic spectrum between the smart meter and the distribution transformer is calculated by using the spectral correlation analysis method, and then the user's transformer area and phase are judged by comparing the spectral correlation degree. Finally, a low-voltage distribution network in Nanjing is selected for field test, and the actual test
results show that the proposed method can effectively complete the recognition of user transformer area and phase.
Key words: deep learning; characteristic harmonic spectrum; harmonic spectrum correlation analysis; transformer area recognition
參考文獻(xiàn)
[1] 余貽鑫,欒文鵬. 智能電網(wǎng)述評(píng)[J]. 中國(guó)電機(jī)工程學(xué)報(bào),2009,29(34):1-8.
[2] 王國(guó)春,楊健. 基于現(xiàn)狀分析的智能用電技術(shù)發(fā)展趨勢(shì)研究[J]. 陜西電力,2016,44(12):49-51.
[3] 李建,趙漢昌. 多功能低壓臺(tái)區(qū)識(shí)別設(shè)備的研制[J]. 電測(cè)與儀表,2014,51(13):107-111.
[4] 袁超. 低壓配電網(wǎng)絡(luò)臺(tái)區(qū)識(shí)別技術(shù)的研究與開(kāi)發(fā)[D]. 大連:大連理工大學(xué),2014.
[5] 阿遼沙·葉,顧君,張小秋. 基于數(shù)據(jù)時(shí)空相關(guān)性的智能臺(tái)區(qū)識(shí)別技術(shù)[J]. 現(xiàn)代電子技術(shù),2018,41(4):30-33.
[6] 潘明明,田世明,吳博,等. 基于智能電表數(shù)據(jù)的臺(tái)區(qū)識(shí)別與竊電檢測(cè)方法研究[J]. 智慧電力,2017,45(12):80-84.
[7] 趙春華,胡恒星,陳保家,等. 基于深度學(xué)習(xí)特征提取和WOA-SVM狀態(tài)識(shí)別的軸承故障診斷[J]. 振動(dòng)與沖擊,2019,38(10):31-37.
[8] HINTON G E, SALAKHUTDINOV R R.Reducing the dimensionality of data with neural network[J].Science,2006,313(5786):504-507.
[9] 趙光權(quán),葛強(qiáng)強(qiáng),劉小勇,等. 基于DBN的故障特征提取及診斷方法研究[J]. 儀器儀表學(xué)報(bào),2016,37(9):1946-1953.
[10] WOLD S.Principal component analysis[J]. Chemometrics & Intelligent Laboratory System,1987,2(1):37-52.
[11] 黃超藝,楊贊鋒,吳榮福,等. 采用皮爾遜相關(guān)性的故障區(qū)段定位新方法[J]. 電氣技術(shù),2018,19(6):56-60.