在電力供應系統中,電力變壓器一直是最受重視的組件之一,若變壓器在運轉期間由於故障造成電力中斷,對經濟及電力供應之穩定性將造成極大的衝擊及影響。因此,變壓器在電力系統中之資產管理已成為企業的營運要素,並著重於提前預防與即時診斷維護,以避免突發事故的發生。其中,絕緣紙的劣化可用於判斷變壓器之耗損程度,絕緣紙嚴重劣化甚至會導致變壓器爆炸,所以藉由絕緣紙之劣化狀況,計算變壓器之剩餘壽命,進而決定變壓器之更換替時機顯得極為重要。 本研究以油浸式變壓器為例,提出一個以即時監控資訊為基之變壓器智慧維護保養建議平台,其目的為協助企業資產管理人員快速的彙整監控設備資訊及採樣資料成報表,給予設備維護診斷建議及剩餘使用壽命評估,以降低突發事故發生機率。本平台採用國際電子電機工程師學會 (Institute of Electrical and Electronics Engineers, IEEE) 所修正之Doernenburg診斷法、Rogers診斷法、以及國際電工委員會 (International Electrotechnical Commission, IEC) 文件中所描述之Duval Triangle診斷法等三種油中氣體分析法,診斷變壓器內部可能的故障,並提供相應的維護對策。而針對壽期預測提出一以資料群集處理技術為基的壽期減損評估方法,以油中氣體及糠醛生成量為變數,建立一個新的壽命評估模型,用以預測變壓器壽命損耗,並以此模型預測結果與IEC壽期減損的評估結果,相互比較及參考。
In the power supply system, the transformer has been one of the most important components. If the transformer faults during the power supply, the economy and stability of the power supply will suffer huge impact and influence. Therefore, the transformer has become a critical factor for business operations in the power system asset management. It focuses on the real-time diagnosis and preventive maintenance to avoid unexpected accidents. In addition, the deterioration of insulating paper can be used to determine the degree of transformer depletion. Serious deterioration of the insulating paper would cause the transformer explosion, and the replacement price of the paper is equivalent to buy a new transformer. Therefore, using the degradation degree of the insulating paper to compute the remaining life of the transformer and determine time for replace or purchase a transformer becomes extremely important. This research is for oil-immersed transformers case study, to propose an intelligence maintenance recommendation platform base on real-time monitoring information. The purpose is to help asset manager quickly compiled the data that collected by the real-time monitoring equipment and regular sampling into information reports. These reports not only give the diagnostic and maintenance advice to engineer but also evaluate equipment remaining life for reduces the incidence of unexpected events. The platform using dissolved gas-in-oil analysis (DGA) base on the three methods which is Doernenburg, Rogers (revised by Institute of Electrical and Electronics Engineers, IEEE) and Duval Triangle (described in the International Electrotechnical Commission (IEC) appendix document) to diagnose transformer internal potential, and provide Maintenance recommend. On the other side, this research proposed a prediction method base on Group method of data handling to estimate life of impairment. This algorithm let the dissolved gas-in-oil and furfural formation to be the input variables to form a new life of impairment assessment model. Using the new proposed model predicted result and another life prediction model proposed by IEC provided for the manager to evaluate life of impairment, compared with each other and reference.