Robust State Estimation And Bad Measurement Identification Of Distribution Network

Along with the application analysis of power system,the accuracy of the underlying data becomes particularly important,and the assessment index of state estimation qualification rate of power grid companies is also attached more and more importance.However,at present,the EMS system still lacks a dedicated detection software for the bad measurement in the network.Undoubtedly,the existence of bad measurements will inevitably affect the accuracy of state estimation and the basic database.Once a large number of bad measurements appear in the basic data,it is likely that the advanced decision analysis of the EMS system will fail,and it will more likely affect the safe operation of the grid.For this reason,this article has made corresponding research from the perspectives of robustness and identification for the existence of bad measurement in the power grid.The second chapter of this paper introduces the basic theory of state estimation from the perspective of robustness and identification.Mainly related concepts and mathematical models.In terms of robustness,the weight function is taken as an example to analyze the robustness of algorithm,which lays the foundation for the third chapter of this paper.In terms of identification,weighted residuals,regularized residuals,and residual ranking search methods are introduced.The principle of maximal regularized residuals testing is emphasized,laying a foundation for the fourth chapter of this paper.In the third chapter,without rejecting the bad measurement,the robust estimation algorithm is directly used to improve the state estimation pass rate and prevent the state estimation pass rate from falling in abnormal situations.This paper proposes an improved state estimation algorithm for exponential weight functions.Artificial operator c is used to combine the last residual and the current residual to construct a new weight function.It can reduce the number of iterations in the algorithm.Meanwhile,it is can be remained that large residuals correspond to small weights and small residuals correspond to large weights.At the same time,due to the effect of the exponential difference,the difference between the residual and the residual becomes larger.the weight between the bad measurement and the normal measurement is enlarged.The bad measurements can be weaken,then the identification accuracy of the robust algorithm is improved.The fourth chapter aims at the measurement existing in the network,and identifies the bad measurement stored in the network one by one from the perspective of identification.This paper proposes a multiple bad measurement identification algorithm based on the residual covariance matrix.The focus of this algorithm is to group the active and reactive power measurements in the distribution network one by one.The grouping of active and reactive power is mainly based on the introduction of a unit complex normalization method and a decoupling state estimation algorithm.The internal measurement grouping of pure active or reactive power measurement is mainly based on the diagonal and off-diagonal elements in the residual covariance matrix to define the correlation coefficient between residual and residual.Use the correlation between residuals to achieve grouping between measurements.Finally,the threshold criterion is used to screen the suspicious measurements and group the suspicious measurements.The measurement corresponding to the largest regularized residual in each group is recorded as a bad measurement.The thesis studies from two perspectives of resistance and identification.The proposed robustness algorithm accelerates the calculation speed of the original algorithm.The proposed identification algorithm improves the efficiency of identifying bad measurements.Both are conducive to the improvement of data quality in energy management systems.