肖耀輝 余俊松 李為明 王玉峰 王永平 薛海平 黃鍇 姚金明



摘 要:由于特高壓換流站系統(tǒng)數(shù)據(jù)來源廣泛、采集密度高、量測裝置多樣、通信協(xié)議復(fù)雜,現(xiàn)有技術(shù)難以對(duì)換流站復(fù)雜狀態(tài)及其隱含的故障特征進(jìn)行準(zhǔn)確辨識(shí)。因此本文提出了基于信息物理融合的特高壓換流站特征識(shí)別技術(shù),在對(duì)以圖像為主的多源異構(gòu)數(shù)據(jù)進(jìn)行預(yù)處理與關(guān)聯(lián)分析后,基于信息物理雙側(cè)狀態(tài)運(yùn)行及遷移特征關(guān)聯(lián)矩陣,對(duì)換流站物理與通信雙側(cè)故障進(jìn)行分析、訓(xùn)練與識(shí)別?;趯?shí)際換流站監(jiān)控圖像進(jìn)行了實(shí)例分析和方法對(duì)比,結(jié)果表明該方法優(yōu)于一些傳統(tǒng)的故障診斷與特征識(shí)別算法,具有較好的診斷能力。
關(guān)鍵詞:大數(shù)據(jù)挖掘;特高壓直流;多源信息融合;信息物理系統(tǒng);故障辨識(shí)
DOI:10.15938/j.jhust.2024.01.008
中圖分類號(hào): TM743? 文獻(xiàn)標(biāo)志碼: A
文章編號(hào): 1007-2683(2024)01-0069-09
EHV Converter Station State Identification Technology Based on Cyber-physical Integration
XIAO Yaohui1, YU Junsong1, LI Weiming1, WANG Yufeng1,WANG Yongping2, XUE Haiping2, HUANG Kai2, YAO Jinming3
(1Maintenance and Test Center of CSG EHV Power Transmission Company, China Southern Power Grid Co, Ltd,?Guangzhou 510000, China;2NR Electric Co, Ltd, Nanjing 210023, China;3College of Automation & College of Artificial Intelligence, Nanjing University of Posts and Telecommunications, Nanjing 210023, China)
Abstract:Due to the wide range of data sources, high collection density, diverse measurement devices and complex communication protocols in the EHV converter station system, it is difficult for the existing technology to accurately identify the complex states of the converter station and its implied fault characteristics Therefore, this paper proposes an information-physical fusion-based feature identification technology for EHV converter stations After pre-processing and correlation analysis of image-based multi-source heterogeneous data, the analysis, training and identification of physical and communication faults of converter stations are performed based on the cyber-physical dual-side state operation and migration feature correlation matrix Example analysis and method comparison based on actual converter station monitoring images are conducted, and the results show that the method outperforms some traditional fault diagnosis and feature identification algorithms and has better diagnostic capability
Keywords:big data mining; UHVDC; multi-source information fusion; information physical system; fault identification
0 引 言
隨著特高壓直流跨區(qū)域傳輸線路的建設(shè)與發(fā)展,我國清潔能源的消納與利用程度得到了大幅度加強(qiáng)。由于特高壓換流站起著連接發(fā)電端和輸配電端的核心作用,保證其安全穩(wěn)定運(yùn)行非常重要。傳統(tǒng)的檢測工作多以人工為主,存在勞動(dòng)強(qiáng)度大的局限性,無法滿足特高壓工程快速發(fā)展的建設(shè)需要[1-2]。
中國目前的電力設(shè)備維護(hù)體系是一種簡單的基于時(shí)間的預(yù)防性試驗(yàn)和定期維護(hù)體系[3]。這種維護(hù)制度由于自身的盲目性,可能會(huì)對(duì)運(yùn)行中的設(shè)備良好運(yùn)行狀態(tài)造成損害[4-5]。
特高壓換流站結(jié)構(gòu)復(fù)雜,具有多電壓交叉特性。雖然在關(guān)鍵設(shè)備與產(chǎn)品的設(shè)計(jì)和生產(chǎn)過程中,整體的絕緣特性得到了可靠的保證,但隨著長期的運(yùn)行,元件的性能退化,運(yùn)行環(huán)境的變化,特別是滲水等要素對(duì)換流站的可靠運(yùn)行造成了潛在的威脅[6-7]?!?br>