Research On Innovation Graph Application For State Estimation

With the rapid development of power system, the Inter-regional, National Supergrid has been formed and its framework and the operation mode are trending complexity. So the fundamental effect of the state estimation becomes more and more prominent, and its accuracy and real-time have a great important meaning for the safety, stability and economy of the power system operation. However, there are still some insufficiency in identifying topology error and bad data on state estimation of power system at present. The Innovation Graph Technique is very good at real-time, so this paper goes deep into the research that the identification of topology change, bad data utilizing Innovation Graph Technique and its practicality in order to expand the theory of Innovation Graph Technique and supply a approach which can identify bad data and topology change in real lager-scale power network quickly and effectively.The model of the Innovation Graph is further studied in this paper. The new model of innovation network is established based on breaker model, The identification of bus topology is converted into the identification of branch topology utilizing breaker model, and it can identify both bus topology change and branch topology change.The model foundation of identifying topology change is unified. The disadvantage of using the detailed physical model to identify topology changes is avoided in the method, and it owns some advantages such as simple in calculation, fast in speed, and much lower requirement in measurement redundancy, and it also can identify bus topology error and branch topology change overlaped together .The method that identifies bad data utilizing the criterion of innovation difference vector is presented based on the Innovation network modeled. The correlation of measurement data and classification mode of bad data in the Innovation Graph State Estimation is presented utilizing innovation difference vector. It is pointed out that identifying bad data located on the link is dominant of identifying bad data in innovation graph technique. The basic character indicated from the innovation difference vector when the different type of bad data exist is analyzed and is focused to identify bad data located on the link. Therefore, the problem whether is necessary to change the tree and how to change it during the process of identifying the bad data based on the Innovation Graph Technique is resolved. It supplies a theory foundation for removing bad data in the measurement system exactly utilizing the Innovation Graph Technique.The application problem of using the sum of loop innovation angle (SLIA) in the Innovation Graph Technique to identify bad data is researched in this paper. The criterion of SLIA which is independed on the criterion of innovation difference vector is constructed, and is used to identify bad data in the detected loop and the strongly relevant bad data which the criterion of the innovation difference vector can't identify. Based on the criterion of SLIA and the criterion of innovation difference vector, the detectability and identifiability of bad data in innovation graph technique is analysed. The ability and efficiency of identifying bad data utilizing the Innovation Graph Technique is improved.Basis on the theoretical analysis mentioned above, the applications of the Innovation Graph state estimation in large-scale electric network are discussed in this paper. Firstly, the influence of model error and arithmetic error to the Innovation Graph Technique is analysed and the rule of optimization is put forward. The optimize spanning tree in the Innovation Graph Technique utilizing physical restriction of power system and the restriction of Innovation Graph Technique itself is established to reduce the disturbance of error . Then the theoretical difference between the innovation graph state estimation and static state estimation and their advantage and disadvantage are analyzed, and the double channel state estimation method is presented. The method uses the Innovation Graph Technique to identify bad data, combines the Innovation Graph Technique and network configuration and utilizes static state estimation to get system state. This enhances the correctness of network topology and the quality of state estimation. Finally, the Innovation Graph Partition Technique is presented according to regional management of large-scale network system. The method makes that the innovation graph state estimation can be carried out respectively in each region, and that bad data or topology change can be identified in each region quickly and effectively. The method is convenient for the security operation and management. The proposed methods are testified in the large-scale electric network, and the results of real topology error identification indicate its feasibility and effectiveness.This research work is supported by the National Nature Science Foundation of China under Grant 50177006.