Researches On Process-Oriented State Estimation Based On PMU's Measurement Information

Power system state estimation is an important part of Energy Management System (EMS), it can provide real-time state for the on-line analysis and control function. Its performance influences the validity of analysis and the effect of control. Traditionally state estimation uses the remote measurements provided by the Supervisory Control And Data Acquisition(SCADA)system, obtaining the state vector iteratively. Nowadays the Phasor Measurement Unit( PMU)becomes an important source of the measurements gradually, compared with SCADA it has the virtue of high precision, strict synchronization of the whole system, short renewal cycle, etc., and can realize the direct measure of the state vectors. Thus the researches concerned with the using of information provided by PMU in the state estimation area have important theoretical and practical meanings.By utilizing PMUs'measurements of partial nodes as well as current phasors construct the innovation network graph under the AC load flow model. The innovation network graph under the AC load flow model is the physical carrier of current innovation. Different from the construction of DC load flow model, the nodal injecting current are either measured directly by PMU or calculated in an indirect way, so these injecting innovations should be kept during the whole process of constructing the innovation network. Choose the branch which had PMU current measurement as link, others as tree, then use link to deduce the innovation of the tree which had no PMU measurements. During the calculation, the effects of nodal injecting innovation source should be taken into account, the branch between innovation injecting source node and ground node should be looked as link and modified again so as to obtain more precise link deduction innovation.When there are ill-conditions such as multiple network configuration changes, the identification logics and estimation calculation procedure are provided. After the construction of the innovation network graph in the AC load flow model, the branch current phasors can be reckoned by the current phasors of the link branch. Aim at the multiple topology changes of the network, the corrected current phasor and the ratio of the corrected current to the forecasting current phasor of the AC load flow model are defined to identify it. After eliminating the influence of the topology changes, by using direct PMU measurements, the corrected current phasors as well as the network parameters the network state can be calculated linearly. The problem of the possible existing PMU bad data is also discussed.A novel process-oriented characteristic state analysis method is put forward, so as to deal with the abundant PMU measurements. First the time process is divided by the innovation network graph in the AC load flow model, then the extreme innovation network graph and the expected innovation network graph, which present the extreme running condition and the medial running condition of the power system, is constructed. Through the computation of these innovation networks, the typical network state representing the running condition of the process can be obtained quickly, and it can provide effective information for the control center to formulate timely and comprehensive strategies.A SCADA/PMU mixed measurements state estimation method based on the process innovation vectors is also put forward. First the PMU current measurements is transformed into power measurements, then mixed with the SCADA measurements and carry out non-linear calculation at the beginning time, thus the sensitivity matrix of the estimated state vector to the power measurements can be obtained. In the following PMU sampling points, the transformed PMU measurements and the pseudo-measurements derived from the load forecasting are mingled together to form the mixed-measurements, then based on obtained sensitivity matrix, the linear tracking state estimation can be carried out in accordance with the sampling cycle of PMU. When the estimation error is accumulated to some extent, the non-linear state estimation is performed once again so as to update the sensitivity matrix.