基于數據-特征增強與參數優化的變壓器故障識別方法
童宇軒1,李燦2,周佳炎1
(1 國網浙江省電力有限公司慈溪市供電公司,浙江 慈溪 315300; 2 浙江省送變電工程有限公司,浙江 杭州 310020)
摘 要 :針對油浸式電力變壓器故障診斷中存在的樣本不平衡及模型參數優化問題,提出一種基于數據-特征增強與自適應參數優化的油浸式變壓器機器學習故障識別模型。通過合成少數類過采樣技術 (SMOTE) 解決數據集故障類別不平衡問題,并結合IEC三比值法構建多維故障特征增強數據集特征表征能力;采用融合正余弦策略和柯西變異機制的改進麻雀搜索算法 (SCSSA) 實現對機器學習模型的自適應調參,來有效提升最小二乘支持向量機 (LSSVM) 超參數尋優性能。實驗對比表明,所提模型較傳統模型具有更高的診斷精度和穩定性,對提升超期服役變壓器的故障診斷能力具有工程應用價值。
關鍵詞 : 變壓器 ;故障診斷;合成少數類過采樣技術;改進麻雀搜索算法;最小二乘支持向量機
中圖分類號 :TM407 文獻標識碼 :B 文章編號 :1007-3175(2025)12-0042-06
Transformer Fault Recognition Method Based on Data-Feature Enhancement and Parameter Optimization
TONG Yu-xuan1 , LI Can2 , ZHOU Jia-yan1
(1 Cixi Power Supply Company of State Grid Zhejiang Electric Power Co., Ltd, Cixi 315300, China; 2 Zhejiang Power Transmission and Transformation Engineering Co., Ltd, Hangzhou 310020, China)
Abstract: Aiming at the problems of sample imbalance and model parameter optimization existing in the fault diagnosis of oil-immersed power transformers, a machine learning fault recognition model for oil-immersed transformers based on data-feature enhancement and adaptive parameter optimization is proposed. The synthetic minority over-sampling technique (SMOTE) is employed to solve imbalance problem in datasets fault categories, and the IEC three-ratio method is combined to construct a multidimensional fault feature set, thereby enhancing the feature representation capability of the dataset. Subsequently, an improved sparrow search algorithm (SCSSA) that integrates sine and cosine strategies and Cauchy mutation mechanisms is adopted to achieve adaptive parameter tuning of machine learning models, effectively enhancing the hyperparameter optimization performance of least squares support vector machines (LSSVM).Comparative experiments demonstrate that the proposed model has higher diagnostic accuracy and stability than the traditional model, and has engineering application value for improving the fault diagnosis capability of transformers that have exceeded their service life.
Key words: transformer; fault diagnosis; synthetic minority over-sampling technique; improved sparrow search algorithm; least squares support vector machines
參考文獻
[1] 任志勇,張天旭 . 一起110 kV 變壓器內部放電故障分析 [J]. 電工電氣,2024(1):71-73.
[2] 謝樂,楊浙,潘成南 . 基于分步特征選取和WOA-LSSVM的變壓器故障診斷 [J]. 電工電氣,2024(8):31-36.
[3] 閔永智,令世文,王果 . 基于混合特征選擇和IOMA-CNN的變壓器故障診斷 [J]. 電力系統保護與控制, 2024,52(23):1-9.
[4] 黨瓏,王琪,董佳寧,等 . 電力變壓器故障診斷方法的比較研究 [J]. 江蘇理工學院學報,2023,29(6) :69-80.
[5] 李雷軍,吳超,付華,等 . 基于油中溶解氣體分析的ISSA優化LSSVM變壓器故障診斷研究 [J]. 電工電能 新技術,2023,42(10):84-94.
[6] 孔博,趙占軍,郭海波,等 . 基于SSA-CNN的電力變壓器 故障智能診斷方法 [J]. 電工技術,2024(23):190-195.
[7] 黃旭,許冬云 . 基于MF-SAE-SSA-KELM油浸式變壓器 故障診斷方法 [J]. 工業控制計算機,2024,37(10): 126-128.
[8] 呂鑫,慕曉冬,張鈞,等 . 混沌麻雀搜索優化算法 [J]. 北京航空航天大學學報,2021,47(8) :1712-1720.
[9] YUAN J H, ZHAO Z W, LIU Y P, et al.DMPPT control of photovoltaic microgrid based on improved sparrow search algorithm[J].IEEE Access,2021,9: 16623-16629.
[10] LIU G Y, SHU C, LIANG Z G, et al.A modified sparrow search algorithm with application in 3d route planning for UAV[J].Sensors,2021,21(4) :1224.
[11] 王玲芝,李晨陽,李程,等 . 多策略改進 SSA優化LSTM網絡的短期光伏發電功率預測 [J]. 武漢大學學 報 ( 工學版 ),2025,58(8) :1356-1366.
[12] 周萱,吳偉麗 . 基于改進 SMOTE不均衡樣本處理和 IHPO-DBN的變壓器故障診斷方法研究 [J]. 電力系統 保護與控制,2024,52(11) :21-30.
[13] 蔡賽男,宋衛星,班利明,等 . 基于鯨魚算法優化 LSSVM的滾動軸承故障診斷 [J]. 控制與決策,2022, 37(1) :230-236.
[14] 劉澤宇,彭澤源,韓愛國 . 基于SCSSA-CNN-BiLSTM的行駛工況下鋰電池壽命預測 [J]. 重慶理工大學學報 ( 自然科學版 ),2024,38(1) :308-318.
[15] 柴巖,常曉萌,任生 . 融合多策略改進的白鯨優化算法 [J]. 計算機工程與應用,2025,61(5) :76-93.