Suzhou Electric Appliance Research Institute
期刊號： CN32-1800/TM| ISSN1007-3175

Article retrieval

文章檢索

首頁 >> 文章檢索 >> 往年索引

基于分步特征選取和WOA-LSSVM的變壓器故障診斷

來源：電工電氣發(fā)布時間：2024-08-30 13:30 瀏覽次數(shù)：135

基于分步特征選取和WOA-LSSVM的變壓器故障診斷

謝樂，楊浙，潘成南
(國網(wǎng)浙江省電力有限公司慈溪市供電公司，浙江 慈溪 315300)
 
    摘 要：為了提高變壓器故障診斷的精度，保障電網(wǎng)的穩(wěn)定運(yùn)行，提出了一種基于 ReliefF 算法與界標(biāo)等距映射(L-Isomap)的分步特征選取和鯨魚群算法(WOA)優(yōu)化最小二乘支持向量機(jī)(LSSVM)的故障診斷模型。選取 7 種常見故障特征油中溶解氣體分析(DGA)氣體以及其構(gòu)造出的16 組比值作為初始特征集，利用 ReliefF 算法分別對初始特征集進(jìn)行特征選擇，再利用 L-Isomap 算法對融合后的特征集進(jìn)行降維處理，將降維處理后的特征集作為故障特征向量代入診斷模型，故障診斷模型采用 WOA-LSSVM 進(jìn)行訓(xùn)練與測試。實驗結(jié)果表明，診斷模型的精度高達(dá)98.31%，相比于其他模型擁有更高的診斷精度。
    關(guān)鍵詞： 變壓器；故障診斷；分步特征選取；降維；鯨魚群算法；最小二乘支持向量機(jī)
    中圖分類號：TM406 ；TM411     文獻(xiàn)標(biāo)識碼：B     文章編號：1007-3175(2024)08-0031-06
 
Transformer Fault Diagnosis Based on Stepwise Feature
Selection and WOA-LSSVM
 
XIE Le, YANG Zhe, PAN Cheng-nan
（Cixi Power Supply Company of State Grid Zhejiang Electric Power Co., Ltd,Cixi 315300, China）
 
    Abstract: In order to improve the accuracy of transformer fault diagnosis and ensure the stable operation of power system. In this paper,proposing a stepwise feature selection based on ReliefF algorithm and landmark isomap (L-Isomap) and a fault diagnosis model for whale optimization algorithm (WOA) least squares support vector machine (LSSVM). The method first selected 7 common fault characteristics dissolved gas analysis in oil (DGA) gas and constructed 16 sets of ratios as the initial feature set. Secondly, the ReliefF algorithm was used to perform feature selection on the initial feature set respectively, and then the L-Isomap algorithm was used to reduce the dimensionality of the fused feature set, and the dimensionality reduction feature set was substituted into the diagnostic model as a fault feature vector, and the fault diagnosis model was trained and tested by WOA-LSSVM. The experimental results show that the accuracy of the diagnostic model is as high as 98.31%, which is higher diagnostic accuracy than that of other models.
    Key words: transformer; fault diagnosis; stepwise feature selection; dimensionality reduction; whale optimization algorithm; least squares support vector machine
 
參考文獻(xiàn)
[1] 劉云鵬，許自強(qiáng)，李剛，等. 人工智能驅(qū)動的數(shù)據(jù)分析技術(shù)在電力變壓器狀態(tài)檢修中的應(yīng)用綜述[J]. 高電壓技術(shù)，2019，45(2) ：337-348.
[2] GUARDADAO J L, NAREDO J L, MORENO P， et al.A comparative study of neural network efficiency in power transformers diagnosis using dissolved gas analysis[J].IEEE Transactions on Power Delivery， 2001， 16(4) ：643-647.
[3] DUVAL M, DEPABLA A.Interpretation of gas-inoil analysis using new IEC publication 60599 and IEC TC10 databases[J].IEEE Electrical Insulation Magazine， 2001， 17(2) ：31-41.
[4] KIM Y M, LEE S J, SEO H D, et al.Development of dissolved gas analysis(DGA) expert system using new diagnostic algorithm for oil-immersed transformers[C]//2012 IEEE International Conference on Condition Monitoring and Diagnosis， 2012 ：365-369.
[5] ROGERS R R.IEEE and IEC codes to interpret incipient faults in transformers， using gas in oil analysis[J].IEEE Transactions on Electrical Insulation， 1978，13(5) ：349-354.
[6] MOLLMANN A, PAHLAVANPOUR B.New guidelines for interpretation of dissolved gas analysis in oilfilled transformers[J].Electra， 1999， 186 ：31-51.
[7] LEE S, KIM Y, SEO H, et al.New methods of DGA diagnosis using IEC TC 10 and related databases Part2:Application of relative content of fault gases[J].IEEE Transactions on Dielectrics and Electrical Insulation， 2013， 20(2) ：691-696.
[8] 謝樂. 基于 DGA 和機(jī)器學(xué)習(xí)的變壓器故障診斷和狀態(tài)預(yù)測研究[D]. 成都：西南交通大學(xué)，2022.
[9] 白星振，臧元，葛磊蛟，等. 變壓器故障診斷用油中溶解氣體征兆優(yōu)選方法[J] . 高電壓技術(shù)，2023，49(9) ：3864-3875.
[10] 謝樂，衡熙丹，劉洋，等. 基于線性判別分析和分步機(jī)器學(xué)習(xí)的變壓器故障診斷[J]. 浙江大學(xué)學(xué)報(工學(xué)版)， 2020，54(11) ：2266-2272.
[11] 李云淏，咸日常，張海強(qiáng)，等. 基于改進(jìn)灰狼算法與最小二乘支持向量機(jī)耦合的電力變壓器故障診斷方法[J].電網(wǎng)技術(shù)，2023，47(4) ：1470-1477.
[12] ZHANG Y, LI X, ZHENG H, et al.A fault diagnosis model of power transformers based on dissolved gas analysis features selection and improved krill herd algorithm optimized support vector machine[J].IEEE Access， 2019， 7 ：102803-102811.
[13] 朱莉，汪小豪，李豪，等. 不平衡樣本下基于變異麻雀搜索算法和改進(jìn) SMOTE 的變壓器故障診斷方法[J].高電壓技術(shù)，2023，49(12) ：4993-5001.
[14] KONONENKO I.Estimating Attributes:Analysis and Extensions of RELIEF[C]//European Conference on Machine Learning， 1994 ：171-182.
[15] DE SILVA V , TENENBAUM J B . Global versus local methods in nonlinear dimensionality reduction[C]//Neural Information Processing Systems， 2002 ：721-728.
[16] MIRJALILI S, LEWIS A.The whale optimization algorithm[J].Advances in Engineering Software，2016， 95(12) ：51-67.
[17] SUYKENS J A K, VANDEWALLE J.Least squares support vector machine classifiers[J].Neural Processing Letters， 1999， 9(3) ：293-300.
[18] 張又文，馮斌，陳頁，等. 基于遺傳算法優(yōu)化 XGBoost 的油浸式變壓器故障診斷方法[J] . 電力自動化設(shè)備，2021，41(2) ：200-206.