Multiinformation-fusion-based Automatic Fault Diagnosis of Smart Substations Protective Relays
In view of low classification accuracy of currently prevailing diagnostic methods,the present work studied a multiinformation-fusion-based automatic fault diagnosis method for protective relays in smart substations.By calculating fault-induced improper actions of line protection according to protection action information,the hidden fault probability model of distance protection was established.According to opening and closing coil currents,fault state characteristics of circuit breakers were extracted by identification ratio.By using Hall current element,the current signals were online mo-nitored and then processed.The characteristic information of different vibration signals was combined,and the information was processed according to specific fusion rules to complete the filling of information in space and time.The decomposition and sequencing of eigenvalues were realized by using M-dimension matrix and according to number of principal compo-nents,respectively,and the fused feature samples were calculated to determine the decision information.A new zero-se-quence direction protection was formed by protecting the measured voltage when a grounding fault occurs.A rule base was built to optimize sample training.According to the propagation mode of the chain fault caused by the hidden fault,the oc-currence of undesired action in adjacent branches was determined to cut off faulty branches,and the power flow crossing limit check was carried out which enabled hidden faults automatic diagnosis by determining state of secondary circuit ac-cording to phase relationship.The proposed method was preliminarily proved to have satisfactory applicative diagnosis util-ity,as it achieved a classification accuracy of 100%,which is the highest among the four groups.