Innovation Graph Based Three-phase State Estimation Of Distribution Network With Distributed Generation

Distribution State Estimation as the foundation of operation and control of thedistribution network is an important means to obtain accurate real-time statusinformation of distribution system. In this paper, the study is expended for the issueof distribution system state estimation with distributed generation, which is based onthe innovation graph theory, then a series of unusual events will be identified on themodel of distribution network innovation graph with distributed generation.The branches close to the root in radical distribution system with distributedgeneration will be influenced by the innovation error, when the abnormal eventsoccur on the injection frequently, which may lead to annihilation or misjudgment ofabnormal events, for which the calculating innovation vector is proposed that is thedifference between the power flow calculated by the injection measurement andinjection prediction. The link-reckoning innovation vector is replaced and the modelof DG is simplified, then the model of distribution network innovation graph withdistributed generation is modeled. At the same time we compare the innovation errorcontains in the two kinds of innovation graph models that is carried out throughMonte Carlo simulation method. The results show that the calculating innovationvector greatly improves the accuracy of the identification of abnormal events.Different from the transmission grid, the injection measurement vector andprediction vector are needed in the conduction of innovation, then the bad dataresults from large measurement error and the sudden load change results from largeprediction error may occur simultaneously, especially in the distribution networkwith DG，the situation is further deteriorated. For the problem that the bad data orrelevant bad data overlap with the load change, a method based on the distributionnetwork innovation graph model with distributed generation is proposed, in whichthe bad data and the load change are separated in the different innovation vector andthe bad data is identified firstly. The example played on the33-buses system showthat the proposed method can accurately identify the overlapping abnormal events.The distribution network is evaluated from the predominantly radial networksinto the weakly meshed networks, in which the contact switch state change willcause a change in topology. For the topology error that the loop closes but not report,we change the identification of the loop into the identification of the equivalentnodal injection by innovation graph theory. The structure of single power supplyradial distribution network is simple which is not conducive to improving reliability.The structure that connected by the switch is widespread, in which the topologychanges frequently. This kind of topology can be identified by the innovation graph theory according to the magnitude and direction of the load transferred. The methodused to identify the topology error is verified in the33-buses system.This research work is supported by the National High Technology Research andDevelopment of China (863Program)(NO.2011AA05A105) and National NaturalScience Foundation of China (NNSFC)(NO.50977017).