电网技术
電網技術
전망기술
POWER SYSTEM TECHNOLOGY
2014年
3期
715-722
,共8页
孙明蔚%童晓阳%刘新宇%甄威%王晓茹
孫明蔚%童曉暘%劉新宇%甄威%王曉茹
손명위%동효양%류신우%견위%왕효여
电网故障诊断%时序贝叶斯知识库%时序约束
電網故障診斷%時序貝葉斯知識庫%時序約束
전망고장진단%시서패협사지식고%시서약속
power system fault diagnosis%temporal Bayesian knowledge bases%temporal constraint
电网故障时有大量报警产生,充分利用报警信号及其时序信息,处理好保护与断路器误动、拒动、信息缺失等不确定性情况,对于电网故障诊断显得非常重要。时序贝叶斯知识库(temporal Bayesian knowledge bases,TBKB)能够清晰表达多个事件之间的时序约束关系,并具备贝叶斯网络的推理能力。建立了基于 TBKB 的电网故障诊断模型,提出了元件故障与保护动作、保护动作与相应断路器跳闸等之间的时序因果关系(TCR)表达、时序约束一致性检查方法。根据电网结构,可先在线搜索出疑似元件,再对它们自动构造TBKB模型。针对信息缺失节点的状态进行假设,形成假设状态组合。针对这些状态组合,通过贝叶斯反向、正向推理,可判断故障元件,误动与拒动的保护与断路器。多个算例验证了该方法的有效性。
電網故障時有大量報警產生,充分利用報警信號及其時序信息,處理好保護與斷路器誤動、拒動、信息缺失等不確定性情況,對于電網故障診斷顯得非常重要。時序貝葉斯知識庫(temporal Bayesian knowledge bases,TBKB)能夠清晰錶達多箇事件之間的時序約束關繫,併具備貝葉斯網絡的推理能力。建立瞭基于 TBKB 的電網故障診斷模型,提齣瞭元件故障與保護動作、保護動作與相應斷路器跳閘等之間的時序因果關繫(TCR)錶達、時序約束一緻性檢查方法。根據電網結構,可先在線搜索齣疑似元件,再對它們自動構造TBKB模型。針對信息缺失節點的狀態進行假設,形成假設狀態組閤。針對這些狀態組閤,通過貝葉斯反嚮、正嚮推理,可判斷故障元件,誤動與拒動的保護與斷路器。多箇算例驗證瞭該方法的有效性。
전망고장시유대량보경산생,충분이용보경신호급기시서신식,처리호보호여단로기오동、거동、신식결실등불학정성정황,대우전망고장진단현득비상중요。시서패협사지식고(temporal Bayesian knowledge bases,TBKB)능구청석표체다개사건지간적시서약속관계,병구비패협사망락적추리능력。건립료기우 TBKB 적전망고장진단모형,제출료원건고장여보호동작、보호동작여상응단로기도갑등지간적시서인과관계(TCR)표체、시서약속일치성검사방법。근거전망결구,가선재선수색출의사원건,재대타문자동구조TBKB모형。침대신식결실절점적상태진행가설,형성가설상태조합。침대저사상태조합,통과패협사반향、정향추리,가판단고장원건,오동여거동적보호여단로기。다개산례험증료해방법적유효성。
After the fault occurs in power grid, many alarm messages are generated. For power system fault diagnosis, it is important to utilize the alarms and their temporal information, and deal with the uncertainty such as mal-function, rejection and incompletion. The theory of temporal Bayesian knowledge bases (TBKB) can clearly express the temporal constraint relationship among multiple events, and possess Bayesian network’s reasoning ability. The TBKB-based power system fault diagnosis models were studied. The expression of temporal casual relationship (TCR) among fault components and protection operations and related breakers tripping are proposed. The consistency checking of TCR was studied, as well as on-line searching algorithm of suspicious components and automatic generating method of TBKB models. For the states of information missing, the state assumption is adopted to create hypothetic state combinations. For these states, Bayesian backward and forward reasoning is made to detect the fault component and identify mal-function and rejection of protections and breakers. The given examples have illustrated that the proposed fault diagnosis method is effective.