Research On Method Of Two-stage State Estimation Considering Network Partition And Sparse Technology

Operation and control of power system are becoming increasingly complex along with its development. To ensure the security and reliability of power system, it is required to get the real-time states speedily, exactly and roundly. But it is hard to obtain the real states exactly for errors of parameters and measurements. State estimation is the most important method to solve this problem. So, this paper researches the aspects of computing speed and accuracy of state estimation. The contents presented in the paper are as follows:At present, the algorithm mostly used in state estimation software is classical weighted least square estimation. It has good estimation quality and convergence performance, but costs a lot of computing time. Considering this problem, making full use of block symmetric and sparse characteristics of information matrix, a fast state estimation method based on orthogonal list of block information matrix is proposed in the paper. Firstly, the power equations are built on the rectangular coordinates in the paper. Secondly, the lower triangular block orthogonal list of block information is created, so that the efficiency of memory space operation and node optimizing code of information matrix is improved. Finally, based on the network node correlative relationship of measurement equations and block information matrix, the measurement equations direct superaddition method of the formation and modification of information matrix is proposed. It avoids the access operation and calculation of jacobian matrix, and improves the efficiency of formation and modification of the block information matrix and the efficiency of the creation of memory space of relevant orthogonal list. On the basis of this algorithm, the power measurement equations are transformed into current measurement equations equivalently. And the part elements of jacobian matrix become constant which is beneficial to modified and improves the calculation efficiency furtherly.Meanwhile, the two-stage state estimation considering the network partition is studied further, aiming at the fact that the bad data in partial areas will contaminate the measure points in other areas. Firstly, by using partition method based on the node degree searching the network is decomposed into radial sub networks, single meshed networks and complex loop networks which are formed many sub areas. Secondly, every sub area is estimated. And according to the areas'estimation results, the weights of measurements are modified. Lastly, the whole network is estimated by using the modified weights. The proposed method inhibits the bad data of partial areas to contaminate other areas'measurements and improves the accuracy of state estimation.The rapidity and validity of the proposed method are verified by the simulation test on IEEE standard examples and practical calculation examples.