Research On The State Estimation Of Mixed Measurement System

At present,along with the development of the national electricity demand and economic,the power system becomes increasingly complex and huge,the requirement of the reliability of the power system is also gradually increasing,need to make better use of the state estimation in order to reflect the state of power grid.As the appearing of the Wide Area Measurement System(WAMS),a new opportunity provide for the traditional state estimation.Compare with traditional measurement system,WAMS system can obtain more accurate measurement data and the phase angle information.But the current WAMS system can not make the entire network be observed,how to coexist with the conventional SCADA measurement for better state estimation work becoming one of the main research hot spots.This paper focus on studying of the state estimation and bad data detection and identification algorithm under the WAMS/SCADA mixed measurement system.Among several kinds of existing mixed measurement state estimation method,the mixed measurement partition state estimation method can obtain an accurate estimation results and ensure a high numerical stability at the same time.But the method doesn't take the zero injection node constraint problem into account,at the same time,the partition and step by step calculation method it taken also brings trouble in the zero injection processing.Focus on this problem,this paper analysing the different nature of measurements and calculation method in the WAMS observable area and unobservable area,proposing a mixed measurement partition state estimation method taking zero injection node constraint into account,the method adopting the matching method to process the zero injection node in the two area,and solving the difficulty in dealing with the zero injection node in partition boundaries,ensuring injection in the zero injection node strictly to 0,and proving the effectiveness of the algorithm in the IEEE system..A complete state estimation procedure not only need state estimation algorithm,but also need effective bad data detection and identification method for bad data detection and judgment,which is an organic whole,neither can do without the other.But traditional bad data detection and identification method can easily affected by the residual submerge,which may lead to the bad data misjudgement.Aiming at the residual submerge problem,this paper proposing a bad data detection and identification scheme based on residual covariance matrix clustering.The scheme clustering the measurements as one class which have great relationships by fuzzy clustering method,in order to providing judgment for the occurrence of the residual submerge.When suspicious data occurring at the measurements with great relationships at the same time,through the distance of suspicious data clustering center and the other measurements,adding the suspicious data set,cooperating with HTI method to judging the measurement where the residual submerge happened and getting the correct identification result.At last,combining the scheme and mixed measurement checking method,forming the bad data identification and detection unit,cooperating with the partition state estimation method taking zero injection node constraint into account,forming a mixed measurement state estimation unit,proving in IEEE39 node system that the method can effectively identify submerged data and eliminate it,correcting bad WAMS measurement data,making the mixing measurement state estimation unit obtaining properly state estimates and ensuring the zero injection node constraint strictly.