- 作者: 張文宇; 楊宏澤; 黃慶連
- 作者服務機構: 成功大學電機工程學系
- 中文摘要: 本文提出一套類神經網路式診斷系統,應用於配電系統之故障判定。經由決策樹之自動推演及決策樹與類神經網路間之直接對映方法,本診斷系統之建立方便而有效。根據配電系統上斷路器與保護電驛的操作狀態,本系統可判定配電系統的故障區間與故障型態。且即使在多重故障或系統保護設備發生錯誤動作或信號傳輸漏失的情形下,本診斷系統仍能提供有效的故障判定。 本文所提出之方法已實際測試於一座實際的台電公司二次變電所;測試結果顯示,本系統無論在線外訓練或線上診斷上均能達到快速而準確的要求。此外,若採用分散式故障信號處理方法,本系統將可有效應用於大型輸配電系統之故障診斷。
- 英文摘要: An on-line fault diagnosis system using a decision tree based neural network is proposed for adistribution system. Through automatic tree induction and direct mapping of the decision tree into amultilayered neural network, the diagnosis system can be designed and implemented with minimum effort.This system can estimate the fault section and the fault type according to the information of protectiverelays and circuit breakers, even in complicated fault situations involving multiple faults, failure or falseoperation of protective devices as well as erroneous or missing transmission signals. The proposed approachhas been practically tested on a typical Taiwan Power (Taipower) secondary substation. The results obtainedshow that extremely fast and accurate processing in both off-line training and on-line diagnosis is achieved.The approach can be well-applied to a transmission and distribution (T&D) system via decentralized controlcenters for on-line and real-time fault diagnosis
- 中文關鍵字: decision tree; layered neural network; fault diagnosis; transmission and distribution
- 英文關鍵字: --